/*
 * EarthCell.
 *
 * The page opens with a full-screen satellite environment and then becomes an ordinary C Level
 * page. This stylesheet does three things and nothing else: it lets the environment fill the
 * window, it puts the writing underneath back into the standard band, and it keeps the site's own
 * header from sitting on top of a map.
 *
 * THE ENVIRONMENT ITSELF IS NOT STYLED HERE, and cannot be. It lives inside a shadow root, which
 * is exactly why it looks the same on this page as it does on a development page: nothing in this
 * file, the theme, or any plugin can reach into it. If the map needs a change, it is in
 * 07-tools/earthcell/widget/src/styles.ts in the growth repository.
 */

/* --------------------------------------------------------------------------
   The environment, full bleed.

   `100vw` INCLUDES THE VERTICAL SCROLLBAR, AND THIS PAGE ALWAYS HAS ONE.

   The old rule was plain `100vw` with `overflow-x: clip` to tidy up after it. Clip hid the
   symptom and kept the fault: the environment was drawn about 15px wider than the window and
   shifted half that to the left, so everything anchored to its right edge was cut off by the
   scrollbar. That included the zoom buttons and the Esri imagery credit, and showing the credit
   is the condition the imagery is used on. Measured on a 1536px window, 28 August 2026.

   `--ec-scrollbar` is the width of that scrollbar. On the root element `100%` is the client
   width, which excludes the scrollbar, while `100vw` includes it, so the difference between them
   is the scrollbar itself. It comes out as 0 on a phone and on any browser with overlay
   scrollbars, which is correct: there is nothing to subtract.

   The clip stays, as a guard against anything else on the page reaching past the edge.
   -------------------------------------------------------------------------- */

:root {
	--ec-scrollbar: calc(100vw - 100%);
}

/* --------------------------------------------------------------------------
   THE WIDTH SYSTEM. ONE BAND, ONE READING COLUMN, AND A NARROWER CAP FOR HEADINGS.

   Set by the Valezim design handoff of 4 September 2026, and Piers's decision of the same day is
   that Kogi and Valezim move to it together. So these are four tokens rather than a Valezim branch,
   and every rule below names a token instead of repeating a number.

   WHAT IT REPLACES. The lower page used four measures, 800 for running text, 660 for a heading and
   its body, 380 for a caveat beside prose and 300 for the readouts, and the chapters written after
   those added 620, 700, 760 and 820 on top. Nine widths is the same as no width: a reader going
   down the page met a new right-hand edge every second block. The band itself was 1120 with 32px
   gutters, and the chapter titles ran to whatever their column left them.

   --ec-band AND --ec-gutter ARE THE ONE CONTAINER. Every section, every chapter inner block and
   every full-bleed band's contents use both, so every element on the page shares one left edge.
   The full-bleed sections still escape the band to paint their ground, and that is not a breakout:
   the words inside them come straight back to these two values, which is what keeps the edge.

   --ec-measure IS THE READING COLUMN. Prose, lists, figures and captions.

   --ec-heading IS NARROWER THAN THE PROSE ON PURPOSE, so a long title breaks to a second line at an
   edge inside the body text rather than overrunning it. 600px IS THE DESIGNER'S CHOICE AND NOT A
   MEASURED RULE. The handoff cites the theme's token audit for it, and the audit does not say that:
   it asks for 30ch on an h2, which at the size it measures works out at roughly half of 600. The
   number may well be right for these headings. It is a preference, so treat it as one.

   The gutter is 24px at every width, including the phone, because the design has no breakpoints and
   a gutter that steps is a left edge that moves.
   -------------------------------------------------------------------------- */

:root {
	/* 1240px since 27 October 2026, Piers's unity pass: the site's own column (header, footer, home
	   page). It was 1040, which set the project pages 100px in from the logo and the menu on each side. */
	--ec-band: 1240px;
	--ec-gutter: 24px;
	--ec-measure: min(680px, 100%);
	--ec-heading: min(600px, 100%);
}

.earthcell-article {
	overflow-x: clip;
}

/*
 * THE WHITE LINE BETWEEN THE MAP AND THE FIRST SECTION, AND WHY IT IS FIXED HERE AND NOT WITH A
 * BACKGROUND COLOUR.
 *
 * Piers, 31 August 2026: "there is some kind of white line between the first and second blocks."
 * There was, twenty pixels of it, the full width of the window.
 *
 * WHERE IT COMES FROM. The page body documents itself in HTML comments, and WordPress wraps a
 * standalone comment in a paragraph. Those paragraphs draw nothing and are zero pixels tall, but
 * they each carry a bottom margin, and the margins collapse into one twenty pixel gap above the
 * first section. Nothing paints that gap, so the site's own near-white background showed through.
 *
 * **THE FIRST FIX WAS TO PAINT THE WHOLE ARTICLE DARK, AND IT WAS WRONG.** This page alternates:
 * some sections paint their own dark background and some are meant to sit on the site's light one,
 * with dark text on them. Painting everything dark left the site cards as near-black text on
 * near-black, which Piers saw within the hour — "you can't see what's going on now". Removing the
 * gap is the fix; hiding it under a colour was not.
 *
 * `:empty` alone does not reach all of them, because a paragraph holding a comment is not empty.
 * Hence the three selectors: the blank ones, the ones followed by a blank one, and the one that
 * sits immediately above a section. Margins collapse through zero-height boxes, so every
 * paragraph in the run has to be neutralised or the gap survives on the one that was missed.
 */
.earthcell-content > p:empty,
.earthcell-content > p:has(+ p:empty),
.earthcell-content > p:has(+ section) {
	margin: 0;
}

.earthcell-article .ec-full {
	width: calc(100vw - var(--ec-scrollbar));
	margin-left: calc(50% - (100vw - var(--ec-scrollbar)) / 2);
	background: #05080b;
}

.earthcell-article clevel-earthcell {
	display: block;
	width: 100%;
}

/*
 * BEFORE THE COMPONENT LOADS there is an empty element on a white page, and the first thing a
 * visitor sees on a slow connection is a blank screen with a heading under it. This gives the
 * space its ground colour immediately, so the page opens dark and the imagery arrives into it.
 */
.earthcell-article clevel-earthcell:not(:defined) {
	display: block;
	min-height: 100vh;
	min-height: 100svh;
	background: #05080b;
}

/* --------------------------------------------------------------------------
   The writing underneath.
   -------------------------------------------------------------------------- */

/* --------------------------------------------------------------------------
   THE HERO. The proposition first, in words, and the map second.

   IT USED TO BE THE OTHER WAY ROUND. A visitor met a full screen of satellite imagery with six
   panels on it, scrolled, and found the only sentence that says what the page is about sitting in
   a white strip between the map and the next dark chapter. Piers reported it on 29 August 2026 as
   a second title in a weird block, and set the order the same day: the Kogi and EarthCell
   proposition, two short paragraphs, then straight into the map.

   IT IS DARK, AND IT RUNS TO THE EDGES, so the page opens on the instrument's own ground and the
   map below it is a continuation rather than a second thing. On paper it read as an interstitial
   strip between two dark pictures, which is exactly what he objected to.

   THE BOTTOM PADDING IS SMALL ON PURPOSE. The map begins immediately underneath and the two are
   meant to read as one opening, not as a band and then a picture.
   -------------------------------------------------------------------------- */

/*
 * THE PAGE'S HEADING IS READ ALOUD AND NEVER DRAWN.
 *
 * The template prints the real h1 here in plain markup so a search engine finds it, and the map
 * instrument prints the same words in its own bar. Only one of the two may be visible or the page
 * shows its title twice, which is the fault this whole change removes.
 *
 * The theme's own screen-reader class already does this and this rule repeats it on purpose. If
 * that class is ever renamed, the heading would otherwise appear as a bare line of text above the
 * map and nobody would connect it to a change in a different stylesheet.
 */
.earthcell-article > .screen-reader-text {
	position: absolute;
	width: 1px;
	height: 1px;
	margin: -1px;
	padding: 0;
	overflow: hidden;
	clip-path: inset(50%);
	white-space: nowrap;
	border: 0;
}

/*
 * THE RULES BELOW STYLE A PANEL THE TEMPLATE NO LONGER PRINTS, and they are kept rather than
 * deleted on 30 August 2026 for one reason: pulling eighty lines out of the middle of this file in
 * the same change that rebuilt the hero would mean re-checking every section on the page to prove
 * nothing shared a rule. That check is worth doing. It is not worth doing in the same breath.
 *
 * Nothing matches .earthcell-head any more, so they draw nothing. They are on the removal list in
 * the dead-rules note further down this file.
 */
.earthcell-article .earthcell-head {
	width: calc(100vw - var(--ec-scrollbar));
	margin-left: calc(50% - (100vw - var(--ec-scrollbar)) / 2);
	background: #05080b;
	color: #e6eef0;
}

.earthcell-article .earthcell-head > * {
	max-width: 1120px;
	margin-left: auto;
	margin-right: auto;
	padding-left: clamp(20px, 5vw, 32px);
	padding-right: clamp(20px, 5vw, 32px);
}

.earthcell-article .earthcell-head > h1 {
	padding-top: clamp(56px, 9vw, 96px);
}

.earthcell-article .earthcell-head > .earthcell-standfirst {
	padding-bottom: clamp(48px, 7vw, 72px);
}

/*
 * THE PAGE'S OWN HEADLINE WAS THE WEAKEST DISPLAY TYPE ON IT.
 *
 * Measured on staging on 29 August 2026: 45.28px at weight 600 in the body-copy colour, against
 * chapter titles at 52 and 54px at weight 800. So "The Heart of the World" read as the most
 * important statement on the page and "Help return Kogi land to Kogi stewardship" read as a
 * caption. Hierarchy comes from contrast, not from a tag name. Reported in the design review the
 * same day.
 *
 * IT NOW OUTRANKS EVERY CHAPTER TITLE: bigger at the top of its clamp, the same weight and the
 * same tracking as they use, in the near-black the instrument uses rather than the grey the
 * paragraphs use. The clamp keeps it from swamping a phone.
 */
.earthcell-head h1 {
	margin: 0;
	font-size: clamp(2.25rem, 5.2vw, 3.5rem);
	line-height: 1.03;
	font-weight: 800;
	letter-spacing: -0.03em;
	/* Reversed out, because the hero band is the instrument's own ground now. */
	color: #f4f8f9;
	max-width: 17ch;
}

/*
 * TWO PARAGRAPHS, NOT ONE SENTENCE. It is a div rather than a p now, because a paragraph cannot
 * legally hold another one, and the two paragraphs inside it are Piers's own from 28 August.
 */
.earthcell-standfirst {
	display: flex;
	flex-direction: column;
	gap: 14px;
	margin: clamp(18px, 2.4vw, 26px) 0 0;
	max-width: none;
}

/*
 * THE QUIET LINK STRAIGHT TO THE ASK. Smaller than the two paragraphs above it and underlined, so
 * it reads as a way out of the hero rather than as a third sentence of the proposition.
 */
.earthcell-standfirst__jump {
	margin-top: 4px;
}

.earthcell-standfirst__jump a {
	font-size: 15px;
	letter-spacing: 0.01em;
	color: var(--ec-lime, #cef17b);
	text-decoration: underline;
	text-underline-offset: 4px;
}

.earthcell-standfirst > p {
	margin: 0;
	max-width: 58ch;
	font-size: clamp(17px, 2.5vw, 19px);
	line-height: 1.6;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content {
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: 0 var(--ec-gutter) 96px;
}

/* The prose itself holds a reading measure. The band is wider than the words on purpose. */
.earthcell-content > p,
.earthcell-content > ul,
.earthcell-content > ol,
.earthcell-content > details {
	max-width: var(--ec-measure);
}

/*
 * A HEADING STOPS SHORT OF ITS OWN BODY TEXT. It was on the same 68ch as the prose, so a two-line
 * title broke wherever the last word happened to fall.
 */
.earthcell-content > h2,
.earthcell-content > h3 {
	max-width: var(--ec-heading);
}

/*
 * EVERY h2 BELOW THE MAP, WHEREVER IT SITS IN THE MARKUP.
 *
 * A page with two heading sizes for the same level reads as two pages. The design handoff puts
 * every h2 below the map at 40px to 48px, weight 800, tight tracking (IMPLEMENTATION-GAPS.md
 * section 0m). The old rule here topped out at 32px, which is the site's article size, and this is
 * not an article.
 *
 * THE SECOND SELECTOR IS NOT DECORATION. The first only matches a heading that is a direct child
 * of the content wrapper, so "A lighter approach to digital monitoring", which sits one level down
 * inside its own section, matched nothing here and fell back to the site's 30px. It was the only
 * heading on the lower page at the wrong size, which is exactly the kind of fault nobody spots
 * because it looks deliberate.
 */
.earthcell-content > h2,
.earthcell-content .ec-section > h2 {
	margin: 88px 0 1.5rem;
	/* Named again here because the second selector is not a direct child, so the cap above misses it. */
	max-width: var(--ec-heading);
	font-size: clamp(28px, 4vw, 40px);
	line-height: 1.1;
	font-weight: 800;
	letter-spacing: -0.03em;
}

/* The folded questions the theme builds from this page's bold paragraphs. */
.earthcell-content > details.pv-faq {
	max-width: var(--ec-measure);
}

.earthcell-content > p,
.earthcell-content > ul,
.earthcell-content > ol {
	font-size: 1.0625rem;
	line-height: 1.65;
}

.earthcell-content > ul,
.earthcell-content > ol {
	padding-left: 1.25rem;
}

.earthcell-content > ul li {
	margin: 0.35rem 0;
}

/*
 * The bracketed notes that are waiting on somebody now have their own treatment further down this
 * file, under "The bracketed notes that are waiting on somebody". The old .earthcell-note class is
 * kept there too, because the theme has always exposed it to page content.
 */

/* --------------------------------------------------------------------------
   The site header.

   It scrolls with the page rather than floating over the environment. A navigation bar on top of
   a satellite instrument reads as two products stacked on each other, and it collides with the
   environment's own bar, which carries the project name.
   -------------------------------------------------------------------------- */

body.page-template-template-earthcell .site-header,
body.page-template-template-earthcell header.site-header,
body.page-template-template-earthcell .main-header {
	position: static;
}

/* --------------------------------------------------------------------------
   The phone.
   -------------------------------------------------------------------------- */

@media (max-width: 780px) {
	.earthcell-head {
		padding: 56px 20px 0;
	}

	.earthcell-content {
		padding: 0 var(--ec-gutter) 64px;
	}

	.earthcell-content > h2 {
		margin: 56px 0 1rem;
	}
}

/* --------------------------------------------------------------------------
   THE TWO BUTTONS THAT GO BACK TO THE MAP.

   BOTH POINT AT #primary, WHICH IS THE TOP OF THE PAGE, and until now the jump was an instant
   snap that landed under the 73px fixed header, so the visitor could not tell the click had done
   anything. Reported in the design review of 29 August 2026.

   THIS STYLESHEET IS ONLY LOADED ON THE EARTHCELL TEMPLATE, which is what makes it safe to set a
   scrolling behaviour on the root element from here. functions.php enqueues it inside an
   is_page_template check.

   THE SMOOTH SCROLL IS OFF FOR ANYBODY WHO ASKS FOR LESS MOVEMENT. A 15,000px animated scroll is
   exactly the kind of motion that setting exists for.
   -------------------------------------------------------------------------- */

@media (prefers-reduced-motion: no-preference) {
	html {
		scroll-behavior: smooth;
	}
}

#primary {
	scroll-margin-top: 80px;
}

/*
 * ANYTHING JUMPED TO BY A LINK STOPS BELOW THE SITE HEADER, NOT BEHIND IT.
 *
 * The C Level header is fixed and painted over the page, so a browser that scrolls an element to
 * the top of the window puts its first seventy-three pixels underneath it. The READ THE PROJECT
 * link at the top of the map points at the first chapter's heading, and the heading was landing
 * behind the header.
 *
 * The environment's own move when the flight lands measures the header and allows for it in code.
 * This is the same allowance for every link that does it the ordinary way.
 */
.earthcell-content [id],
.earthcell-article section[id] {
	scroll-margin-top: 88px;
}

/* ==========================================================================
   THE LOWER PAGE, REBUILT 28 AUGUST 2026.

   THE FAULT THIS SECTION FIXES. Everything below the map ran as one column of paragraphs at the
   full width of the band. IMPLEMENTATION-GAPS.md section 0m names that as the main reason the
   lower page read wrong: a paragraph the full width of the band is not a paragraph anybody
   finishes. The measure inside the band is set separately from the band itself.

   THE FOUR MEASURES THIS SECTION WAS BUILT ON ARE GONE, replaced on 4 September 2026 by the width
   system at the top of this file. They were 800px for running text, 660px for a heading and its
   body, 380px for a caveat beside the prose and 300px for the monitoring readouts. The first two
   are now the one reading column, --ec-measure. The last two are columns of a grid rather than
   passages of prose, so they keep their own widths.

   WHY THE RULES ARE HERE AND NOT ON THE PAGE. The page body is WordPress content, and WPBakery
   removes a style element or a stylesheet link from page content the first time anybody opens the
   page in the builder and saves it. The page then looks broken with nothing in the revision
   history to explain it, which is exactly what happened to the calculator pages in August 2026.
   So the page carries class names and this file carries every rule they refer to.

   THE CLASS NAMES ARE MATCHED IN 07-tools/earthcell/page/wordpress-content.html in the growth
   repository. If a class is renamed there it must be renamed here in the same change, because
   nothing in WordPress will warn that a selector has stopped matching.
   ========================================================================== */

/* --------------------------------------------------------------------------
   THIRTY-EIGHT CLASS NAMES BELOW THIS LINE ARE NO LONGER ON THE PAGE.

   They are the superseded article design: .ec-band and its teal and dark variants, .ec-band__inner,
   .ec-grid--pair, .ec-grid--readouts, .ec-slot, .ec-figure and its two parts, .ec-caveat, .ec-cta
   and its four parts, .ec-contrast and its five, .ec-diagram, .ec-node and its five, .ec-readout
   and its four, .ec-stream and its two, .ec-note, .ec-note--image-slot, .ec-notice and .ec-full.

   THEY ARE LEFT IN ON PURPOSE AND SHOULD COME OUT. Removing four hundred lines of rules at the end
   of the session that rebuilt all ten chapters would have meant re-checking every section on the
   page again to prove nothing had been sharing a rule. That check is worth doing; it is not worth
   doing tired. The list above is the removal, and the way to prove it is the same script that
   found it: read every class attribute out of page/wordpress-content.html and every .ec- selector
   out of this file, and subtract.

   .earthcell-note, without the ec- prefix, is NOT in that list. It is the name the theme has
   always exposed to ordinary page content and something else on the site may still use it.
   -------------------------------------------------------------------------- */

/*
 * The palette, declared once on the content wrapper so every rule below can name a colour rather
 * than repeat a hex. These are the C Level tokens from IMPLEMENTATION-GAPS.md section 11 and no
 * others: the design adds no colour to the site's own set.
 */
.earthcell-content {
	--ec-ink: #3e4454;
	--ec-paper: #f4f8f9;
	--ec-teal: #005a69;
	--ec-lime: #cef17b;
	--ec-aqua: #00eed2;
	--ec-ground: #05080b;
	--ec-mono: ui-monospace, SFMono-Regular, Menlo, Consolas, monospace;
}

/* ==========================================================================
   C1 SEQUENCE. The first chapter below the map.

   THE SECTION IS 300vh TALL AND THAT HEIGHT IS THE CHAPTER. The pane inside it sticks to the top
   of the window and stays there for the whole 300vh, so the visitor scrolls three screens' worth
   while one still frame changes in front of them. If this ever renders as an ordinary section
   with a map that does not move, the chapter is missing rather than merely unstyled.

   NOTHING MAY SET overflow ON THE SECTION ITSELF. A sticky child stops sticking as soon as any
   ancestor scrolls or clips on both axes. The overflow: hidden this chapter needs belongs on the
   pane inside it, which is what the rules below do. The same reason is why .earthcell-article at
   the top of this file uses overflow-x: clip and not overflow: hidden.

   THE FULL-BLEED SUM IS THE ONE .ec-band USES, AND FOR THE SAME REASON. --ec-scrollbar holds the
   tokens "calc(100vw - 100%)" rather than a measured number, and the 100% is worked out where the
   property is finally used. Here that is inside .earthcell-content, which is 1056px at desktop
   rather than the window, so the sum would cancel itself out and the section would stay as wide
   as the prose. A plain 100vw pulled left by half the column and right by half the window lands
   centred on the visible window and overhangs by half a scrollbar at each end, which
   .earthcell-article already clips. Measured cost at the narrowest inset: the 18px gutter becomes
   about 10.5px of visible gutter in a Windows desktop browser under 400px wide, and stays a full
   18px on a phone and in any browser with overlay scrollbars, where there is no scrollbar to
   subtract.

   EVERY WORD OF THIS CHAPTER IS IN THE PAGE, NOT IN THE WIDGET. The map div is deliberately
   empty. A search engine, and a person whose JavaScript never runs, both read the whole chapter
   from the HTML, and the ground colour on the empty map div is what that reader sees behind it.
   That is why every stage rule below is gated on the data-ec-stage attribute: the widget writes
   that attribute, so with no widget there is no attribute, and the page falls back to a lit stage
   list and two visible cards instead of grey text and empty boxes.

   THE ESRI CREDIT IS NOT DECORATION. Showing it is the condition the imagery is used on. The
   chapter map is built with no MapLibre attribution control, because at full screen that control
   lands in the bottom-right corner underneath the record card. This line replaces it, it is the
   last child of the right-hand stack in normal flow at z-index 2, and no scroll stage can hide
   it. Nothing in this chapter may take a z-index above 2, and nothing may be positioned over it.

   SAMPLE, ILLUSTRATIVE AND DEMONSTRATION GEOMETRY STAY. The cell code, the funding claim and the
   three years all come from the demonstration boundary. CLAUDE.md requires each to be marked on
   screen while that is true, and removing the markers would make the chapter say something that
   is not so.

   NO BREAKPOINTS AT ALL. Type and spacing are clamp() and the two rows wrap on flex. There was
   one media query here, stepping a top margin the chapter should never have had, and both went in
   review. A breakpoint written here is a fault, not a refinement.
   ========================================================================== */

/*
 * NO TOP MARGIN, AND NO BREAKPOINT TO STEP IT DOWN. Both were here and both were review faults.
 *
 * The design gives this chapter no margin at all: the map ends and the dark chapter begins. An
 * 88px margin puts a strip of the light page ground between the two, because the article wrapper
 * has no background of its own, so the reader gets a white band across the join. The design gives
 * a top margin only to C3 and C8, and both of those follow a light chapter.
 *
 * The breakpoint went with it. The handoff is explicit: there are no media queries below the map,
 * columns collapse when their content stops fitting, and a breakpoint written here is a fault
 * rather than a refinement. Removing the margin removed the only thing that one was stepping.
 */
.earthcell-content .ec-c1 {
	position: relative;
	width: 100vw;
	margin-left: calc(50% - 50vw);
	height: 300vh;
	background: var(--ec-ground);
	color: #e6eef0;
}

/*
 * The pane that sticks. 100vh, and NOT 100dvh, which is the opposite of the rule everywhere else
 * in this project and is deliberate.
 *
 * 100dvh is the honest height of a phone window and it is right for a panel the reader has to
 * reach. It is wrong for a frame that is meant to hold still. The dynamic height changes every
 * time the address bar slides, so the pinned frame would resize during the very scroll it exists
 * to hold steady, and the map canvas inside it would lay itself out again on each change. The
 * design says 100vh and this is why.
 *
 * The overflow: hidden is here and not on the section above. On the section it would stop its own
 * sticky child from sticking.
 */
.earthcell-content .ec-c1__inner {
	position: sticky;
	top: 0;
	height: 100vh;
	overflow: hidden;
	/*
	 * THE TWO ROWS ARE FLEX ITEMS NOW, NOT TWO ABSOLUTE BLOCKS, AND THIS IS A BUG FIX.
	 *
	 * They used to be pinned one to the top and one to the bottom of this frame, and each grew
	 * inwards. On a window 900px tall they cleared each other, which is the height the design was
	 * drawn at. At 1132 by 700 they overlapped by 75 pixels and at 390 by 667 by 91, so the stage
	 * list printed straight through the headline. Reported in the design review of 29 August 2026
	 * and measured on staging before this change.
	 *
	 * space-between puts the gap between them rather than under them, so the overlap cannot happen
	 * at any height. What is left over on a short screen is a smaller gap, not two blocks of text
	 * on top of each other.
	 */
	display: flex;
	flex-direction: column;
	justify-content: space-between;
}

/*
 * The empty div the widget mounts the map into. The ground colour is what shows before the first
 * tile arrives, and it is the whole of what shows when the widget never loads at all.
 *
 * THE .earthcell-content PREFIX ON THIS RULE IS LOAD-BEARING. MapLibre adds its own
 * .maplibregl-map class to this very div, and maplibre-gl.css sets position: relative on it. That
 * selector scores 0-1-0. This one scores 0-2-0 and wins outright, so the map stays in its
 * inset: 0 frame. Written as a bare .ec-c1__map it would tie at 0-1-0 and the outcome would turn
 * on which stylesheet the browser loaded last.
 */
.earthcell-content .ec-c1__map {
	position: absolute;
	inset: 0;
	z-index: 0;
	background: var(--ec-ground);
}

/*
 * The darkening over the imagery. Heaviest at the top and the bottom, where the words sit, and
 * almost clear across the middle, where the boundary and the cells are. It takes no pointer
 * events because the map underneath it is the thing being looked at.
 */
.earthcell-content .ec-c1__scrim {
	position: absolute;
	inset: 0;
	z-index: 1;
	pointer-events: none;
	/*
	 * THE MIDDLE TWO STOPS WERE 0.1 AND 0.2 AND THE LEAD PARAGRAPH SAT ON THEM.
	 *
	 * The design draws this gradient heavy at the top and the bottom and nearly clear across the
	 * middle, which is right for the boundary and the grid. It is wrong for the words: the lead
	 * paragraph sits at about 60 per cent down, so its real contrast changed with whatever the
	 * satellite happened to show underneath it, and over a pale hillside it dropped below 3 to 1.
	 * Reported in the design review of 29 August 2026.
	 *
	 * 0.38 and 0.48 keep the imagery clearly readable through the middle and put a floor under the
	 * text. The cells and the boundary are drawn in lime and white and are unaffected at these
	 * values; checked on screen at three window sizes.
	 */
	background: linear-gradient(180deg, rgba(5, 8, 11, 0.72) 0%, rgba(5, 8, 11, 0.38) 34%, rgba(5, 8, 11, 0.48) 58%, rgba(5, 8, 11, 0.92) 100%);
}

/* --------------------------------------------------------------------------
   The top row: the eyebrow on the left, the four stage labels on the right.
   -------------------------------------------------------------------------- */

/*
 * 88px AT LEAST, BECAUSE THE SITE'S OWN HEADER IS PAINTED OVER THIS PANE.
 *
 * The design was drawn on a bare browser window and puts this row at clamp(18px, 3.5vw, 40px)
 * from the top. On the real page the header is fixed, 73px tall on a desktop and 68px on a phone,
 * and this pane is pinned at top 0 underneath it. At 40px the eyebrow and the first stage item
 * were behind it and nobody could read either. Measured on staging at 1536 by 900.
 *
 * max() rather than a breakpoint. Below the map there are no media queries at all, and there is no
 * need for one: the design's own value is used wherever it is already clear of the header, and
 * 88px takes over where it is not. It is the same 88px the map chrome above uses, for the same
 * reason, with the same margin for the header growing a line.
 */
.earthcell-content .ec-c1__top {
	/*
	 * RELATIVE, NOT ABSOLUTE. The z-index still has to lift this row over the map and the scrim,
	 * and z-index needs a position, so it is relative rather than static. What was `top` and the
	 * two insets are padding now, because a flex item is placed by its parent.
	 */
	position: relative;
	z-index: 2;
	/*
	 * THE 88px CLEARS THE SITE'S OWN FIXED HEADER, which is 73px on a desktop and 68px on a phone.
	 * It is a clamp on the window height now rather than a constant, so a short screen gives back
	 * the dozen pixels it does not need. It never goes below 76, which still clears the phone
	 * header with room.
	 */
	padding: max(clamp(18px, 3.5vw, 40px), clamp(76px, 11vh, 88px)) clamp(18px, 4.5vw, 48px) 0;
	/*
	 * IT GIVES WAY FIRST ON A SHORT SCREEN. The headline and the two cards are the chapter; the
	 * stage list is the label on it. If one of them has to lose a few pixels on a small phone it
	 * is this one, and clipping the foot of a four-item list is far better than printing it
	 * through the headline, which is what happened before.
	 */
	flex: 0 1 auto;
	min-height: 0;
	overflow: hidden;
	/*
	 * AND IF IT STILL HAS TO CLIP, THE CUT FADES RATHER THAN CHOPS.
	 *
	 * On a screen short enough that the fourth stage will not fit, a hard cut through the middle of
	 * a line of type reads as a rendering fault. A twelve pixel fade at the foot reads as a list
	 * that continues, which is what it is. The values below are tuned so that a 390 by 667 phone
	 * shows all four and the fade never appears; it is the guard for anything shorter.
	 */
	mask-image: linear-gradient(180deg, #000 calc(100% - 12px), transparent 100%);
	display: flex;
	flex-wrap: wrap;
	align-items: flex-start;
	justify-content: space-between;
	gap: 14px 40px;
}

.earthcell-content .ec-c1__eyebrow {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.22em;
	color: #8c9ba3;
}

/*
 * THE THEME'S LIME ARROW HAS TO BE TURNED OFF ON THIS LIST TOO.
 *
 * The site draws a 24px lime disc with an arrow in it on every list item, through
 * .entry-content li::before. The rule further down this file switches it off for the contrast
 * columns and the returns list, and its comment says every list inside .earthcell-content gets
 * its own mark and none of them get the arrow. This is a new list, so it needs its own line: a
 * forward arrow beside each stage label would read as four buttons rather than four steps, and it
 * would push the labels off their own squares.
 *
 * The theme's selector scores 0-1-1. This one scores 0-2-1 and wins.
 */
.earthcell-content .ec-c1__stages li::before {
	content: none;
	display: none;
}

.earthcell-content .ec-c1__stages {
	list-style: none;
	margin: 0;
	padding: 0;
	/*
	 * TWO COLUMNS WHEN ONE WILL NOT FIT, WITHOUT A BREAKPOINT.
	 *
	 * A single column of four items is 204px tall on a phone, and that plus the headline and the
	 * two cards is more than a 667px screen holds. auto-fit puts the four in two columns as soon
	 * as the row is wide enough for two 150px cells, which halves the height on a phone and leaves
	 * the desktop as the design drew it, because at 232px only one column fits.
	 */
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 150px), 1fr));
	gap: clamp(6px, 1.2vh, 9px) 18px;
	min-width: 232px;
}

.earthcell-content .ec-c1__stage {
	margin: 0;
	display: flex;
	align-items: center;
	gap: 12px;
}

/*
 * The square beside each label. It is an outline until the chapter reaches that stage, and it
 * fills only while that stage is the current one, so the visitor can see both how far through the
 * sequence they are and which step they are looking at.
 */
.earthcell-content .ec-c1__swatch {
	flex: none;
	width: 9px;
	height: 9px;
	border: 1px solid rgba(230, 238, 240, 0.3);
	background: transparent;
	transition: background 0.35s ease, border-color 0.35s ease;
}

.earthcell-content .ec-c1__stage-label {
	font-family: var(--ec-mono);
	font-size: 11px;
	/* 1.45 rather than 1.6. Two of these four labels wrap, so the leading is worth eight pixels. */
	line-height: 1.45;
	letter-spacing: 0.1em;
	color: #5e6c74;
	transition: color 0.35s ease;
}

/*
 * REACHED. A stage lights every label at or below its own number, so each of the four items needs
 * its own selector naming the stage values that reach it. The first item is reached by every
 * stage, which is why its selector only asks that the attribute exists.
 *
 * The script writes data-ec-stage on the section and paints nothing itself. Everything the
 * visitor sees change is one of these rules reacting, which is what keeps the whole look of this
 * chapter in this file.
 */
.earthcell-content .ec-c1[data-ec-stage] .ec-c1__stage[data-ec-stage-index="0"] .ec-c1__swatch,
.earthcell-content .ec-c1:is([data-ec-stage="1"], [data-ec-stage="2"], [data-ec-stage="3"]) .ec-c1__stage[data-ec-stage-index="1"] .ec-c1__swatch,
.earthcell-content .ec-c1:is([data-ec-stage="2"], [data-ec-stage="3"]) .ec-c1__stage[data-ec-stage-index="2"] .ec-c1__swatch,
.earthcell-content .ec-c1[data-ec-stage="3"] .ec-c1__stage[data-ec-stage-index="3"] .ec-c1__swatch {
	border-color: var(--ec-lime);
}

.earthcell-content .ec-c1[data-ec-stage] .ec-c1__stage[data-ec-stage-index="0"] .ec-c1__stage-label,
.earthcell-content .ec-c1:is([data-ec-stage="1"], [data-ec-stage="2"], [data-ec-stage="3"]) .ec-c1__stage[data-ec-stage-index="1"] .ec-c1__stage-label,
.earthcell-content .ec-c1:is([data-ec-stage="2"], [data-ec-stage="3"]) .ec-c1__stage[data-ec-stage-index="2"] .ec-c1__stage-label,
.earthcell-content .ec-c1[data-ec-stage="3"] .ec-c1__stage[data-ec-stage-index="3"] .ec-c1__stage-label {
	color: #f4f8f9;
}

/* CURRENT ONLY. Four exact matches, no :is(), because only one square is filled at a time. */
.earthcell-content .ec-c1[data-ec-stage="0"] .ec-c1__stage[data-ec-stage-index="0"] .ec-c1__swatch {
	background: var(--ec-lime);
}

.earthcell-content .ec-c1[data-ec-stage="1"] .ec-c1__stage[data-ec-stage-index="1"] .ec-c1__swatch {
	background: var(--ec-lime);
}

.earthcell-content .ec-c1[data-ec-stage="2"] .ec-c1__stage[data-ec-stage-index="2"] .ec-c1__swatch {
	background: var(--ec-lime);
}

.earthcell-content .ec-c1[data-ec-stage="3"] .ec-c1__stage[data-ec-stage-index="3"] .ec-c1__swatch {
	background: var(--ec-lime);
}

/*
 * NO JAVASCRIPT, NO ATTRIBUTE, AND THE LIST STILL HAS TO BE READABLE. Every rule above waits for
 * data-ec-stage, and nothing writes it if the widget bundle never loads. Left at their base
 * colours the four labels are grey on a grey outline over a dark ground, which is the copy hidden
 * in plain sight. So the no-attribute state is the lit one.
 */
.earthcell-content .ec-c1:not([data-ec-stage]) .ec-c1__swatch {
	border-color: var(--ec-lime);
}

.earthcell-content .ec-c1:not([data-ec-stage]) .ec-c1__stage-label {
	color: #f4f8f9;
}

/* --------------------------------------------------------------------------
   The bottom row: the heading and its sentence on the left, the two cards and the imagery credit
   on the right.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c1__bottom {
	position: relative;
	z-index: 2;
	padding: 0 clamp(18px, 4.5vw, 48px) clamp(24px, 4vh, 52px);
	/* It never shrinks. The headline and the two cards are the chapter. */
	flex: 0 0 auto;
	display: flex;
	flex-wrap: wrap;
	align-items: flex-end;
	justify-content: space-between;
	gap: clamp(16px, 2.6vh, 22px) 48px;
}

.earthcell-content .ec-c1__words {
	max-width: var(--ec-measure);
	display: flex;
	flex-direction: column;
	gap: clamp(12px, 2vh, 18px);
}

/*
 * THE MARGIN: 0 IS NOT THERE TO BEAT ANOTHER RULE, IT IS THERE TO STOP ONE. This section is not
 * .ec-section, so the 88px heading margin further up this file never reaches it. What does reach
 * it is the site's own h2 margin, and inside a flex column that is pinned to the bottom of the
 * pane that margin pushes the heading down over the sentence under it.
 */
.earthcell-content .ec-c1__title {
	margin: 0;
	max-width: none;
	font-size: clamp(29px, 7.2vw, 54px);
	line-height: 1.02;
	font-weight: 800;
	letter-spacing: -0.03em;
	color: #f4f8f9;
}

.earthcell-content .ec-c1__lead {
	margin: 0;
	max-width: 580px;
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.6;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content .ec-c1__cards {
	display: flex;
	flex-direction: column;
	gap: clamp(8px, 1.4vh, 10px);
	min-width: 238px;
}

.earthcell-content .ec-c1__card {
	display: flex;
	flex-direction: column;
	padding: clamp(11px, 1.9vh, 14px) 16px;
	background: rgba(8, 13, 17, 0.8);
	opacity: 1;
	transform: none;
	transition: opacity 0.45s ease, transform 0.45s ease;
}

.earthcell-content .ec-c1__card--support {
	gap: 6px;
	border: 1px solid rgba(206, 241, 123, 0.5);
}

.earthcell-content .ec-c1__card--record {
	gap: 9px;
	border: 1px solid rgba(230, 238, 240, 0.2);
}

/*
 * THE CARDS ARE HIDDEN ONLY ONCE THE ATTRIBUTE EXISTS, and that gate is the whole point. Both
 * cards carry copy that is in the HTML, so a rule that hid them without waiting for
 * data-ec-stage would leave that copy permanently invisible on any page where the widget failed
 * to load, with nobody able to read it and nothing on screen to say why.
 *
 * The visibility is delayed out and instant back so a hidden card cannot be read by a screen
 * reader or reached by a keyboard while it is fading.
 */
.earthcell-content .ec-c1[data-ec-stage] .ec-c1__card {
	opacity: 0;
	visibility: hidden;
	transform: translateY(6px);
	transition: opacity 0.45s ease, transform 0.45s ease, visibility 0s linear 0.45s;
}

/* The support card arrives at the third stage, the record card at the fourth. */
.earthcell-content .ec-c1:is([data-ec-stage="2"], [data-ec-stage="3"]) .ec-c1__card--support,
.earthcell-content .ec-c1[data-ec-stage="3"] .ec-c1__card--record {
	opacity: 1;
	visibility: visible;
	transform: none;
	transition: opacity 0.45s ease, transform 0.45s ease, visibility 0s;
}

.earthcell-content .ec-c1__card-label {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.18em;
}

.earthcell-content .ec-c1__card--support .ec-c1__card-label {
	color: var(--ec-lime);
}

.earthcell-content .ec-c1__card--record .ec-c1__card-label {
	color: #8c9ba3;
}

/*
 * The cell code. The page ships EC-KOGI-000027 so the code is in the HTML for a reader and for a
 * search engine, and the script overwrites it with the code of the cell it actually flew to, so
 * the card can never show an identifier that is not on the map beside it. Either way it is a
 * demonstration code, which is what the SAMPLE marking above it says.
 */
.earthcell-content .ec-c1__cell-code {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 13px;
	line-height: 1.4;
	letter-spacing: 0.06em;
	color: #f4f8f9;
}

/* The three-point timeline: one square done, two to come, with the rules stretching between. */
.earthcell-content .ec-c1__track {
	display: flex;
	align-items: center;
	gap: 8px;
}

.earthcell-content .ec-c1__node {
	flex: none;
	width: 7px;
	height: 7px;
	border: 1px solid rgba(230, 238, 240, 0.4);
	background: transparent;
}

.earthcell-content .ec-c1__node--done {
	background: var(--ec-lime);
	border-color: var(--ec-lime);
}

.earthcell-content .ec-c1__rule {
	flex: 1;
	height: 1px;
	background: rgba(230, 238, 240, 0.28);
}

.earthcell-content .ec-c1__years {
	margin: 0;
	max-width: none;
	display: flex;
	justify-content: space-between;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1.6;
	letter-spacing: 0.1em;
	color: #a8b6be;
}

/*
 * THE IMAGERY CREDIT. It is the last line of the right-hand stack, in normal flow, and it is
 * always on screen because no stage rule touches it. Showing it is the condition Esri's imagery
 * is used on, which is why the chapter map is built with the MapLibre attribution control turned
 * off: at full screen that control lands in the bottom-right corner under the record card and
 * would be covered rather than seen. Do not move this line into a card, do not let a stage hide
 * it, and do not position anything over it.
 */
/*
 * THE 64px OF PADDING ON THE RIGHT IS THE CHAT BUBBLE, and it is the reason this rule is not just
 * a font size.
 *
 * The site runs a Brevo chat widget, which is fixed to the bottom right corner of every page: 60px
 * across, 60px tall, 35px in from the right edge and 20px up from the bottom. This credit is the
 * last line of the right-hand stack, so it ran straight underneath it and the last two words,
 * DEMONSTRATION GEOMETRY, were behind an orange circle. Found with document.elementFromPoint
 * across the width of the line, not by looking at it, on staging on 28 August 2026.
 *
 * Showing the imagery credit is the condition Esri's imagery is used on. It is the one thing on
 * this chapter that is not a preference, so the credit moves and the bubble stays where the rest
 * of the site puts it. 64px clears the bubble's own 95px of occupied width with room to spare, and
 * it is a constant because the bubble is a constant at every window size.
 */
.earthcell-content .ec-c1__credit {
	margin: 0;
	max-width: none;
	padding-right: 64px;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1.6;
	letter-spacing: 0.12em;
	color: #a8b6be;
}

/*
 * MOTION OFF, STORY INTACT. The visitor who asks their machine for less movement still gets every
 * stage: the squares still light, the cards still appear, the map still moves between the four
 * views. What goes is the fading and the sliding, here in the stylesheet, and the camera easing,
 * in the script, which jumps instead.
 */
@media (prefers-reduced-motion: reduce) {
	.earthcell-content .ec-c1__swatch,
	.earthcell-content .ec-c1__stage-label,
	.earthcell-content .ec-c1__card {
		transition: none;
	}

	.earthcell-content .ec-c1[data-ec-stage] .ec-c1__card {
		transform: none;
	}
}

/* --------------------------------------------------------------------------
   Section rhythm.

   88px between bands, 72px between blocks that belong to each other (section 0m). The bands
   themselves do not repeat the band width: they are inside .earthcell-content, which already
   holds it.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-section {
	margin-top: 88px;
}

/*
 * The rule above "Explore Kogi", from section 0B.2. It separates the introduction from the first
 * real section, and the 40px is the gap between the rule and the heading, not the gap between the
 * sections, which is why both a margin and a padding are set here.
 */
.earthcell-content .ec-section--rule {
	padding-top: 40px;
	border-top: 1.5px solid rgba(62, 68, 84, 0.16);
}

.earthcell-content .ec-section--narrow {
	max-width: var(--ec-measure);
}

/*
 * A SECTION THAT FOLLOWS A FULL-BLEED BAND DROPS ITS RULE AND ITS PADDING.
 *
 * The rule and its 40px exist to separate one light section from the next. After a band that runs
 * to both edges of the window there is nothing to separate: the edge of the band has already done
 * it. Leaving both produced a horizontal line sitting in about 130px of empty space, which Piers
 * reported on 29 August 2026 as a huge gap and a line that both needed to go.
 */
.earthcell-content .ec-c1 + .ec-section--rule,
.earthcell-content .ec-c3 + .ec-section--rule,
.earthcell-content .ec-c4 + .ec-section--rule,
.earthcell-content .ec-c5 + .ec-section--rule,
.earthcell-content .ec-c6 + .ec-section--rule,
.earthcell-content .ec-c8 + .ec-section--rule {
	padding-top: 0;
	border-top: 0;
}

/*
 * THE FIRST CHAPTER TOUCHES THE MAP. IT HAS NOTHING ABOVE IT TO BE SEPARATED FROM.
 *
 * THIS IS THE PALE BAND PIERS REPORTED ON 30 AUGUST, the second one, and it is the same fault as
 * the rule below in a place that rule could not reach. Every chapter carries a top margin to hold
 * it off the one above. The first chapter has no chapter above it: it has the map instrument, which
 * is full-bleed and dark and ends flush. So its margin showed 88px of the pale page ground in a
 * band right across the window, directly under the map.
 *
 * MEASURED ON STAGING AT 1440 WIDE: the map ended at 900 and the first chapter began at 988.
 *
 * IT IS first-of-type AND NOT first-child, and that is not a preference. WordPress wraps each HTML
 * comment in the page body in a paragraph tag, so the first child of this wrapper is an empty
 * paragraph and a first-child rule would land on that and do nothing. The chapters are the only
 * sections in here.
 *
 * A CHECK THAT WALKS THE CHAPTERS COMPARING EACH WITH THE ONE ABOVE IT CANNOT SEE THIS. That is how
 * it survived the pass that fixed the other two: the first chapter has no previous sibling, so it
 * was the one element the comparison never tested. If you are verifying spacing on this page, start
 * above the first chapter and finish below the last one.
 */
.earthcell-content > section:first-of-type {
	margin-top: 0;
}

/*
 * TWO DARK CHAPTERS IN A ROW TOUCH. THEY DO NOT GET A PALE STRIPE BETWEEN THEM.
 *
 * THIS IS THE "WHITE EMPTY BLOCKS" THE CLIENT REPORTED ON 30 AUGUST 2026, and the stripe is not
 * white. The site's own page ground is #F4F8F9, and every one of these bands escapes the reading
 * column to run edge to edge, so a top margin on one of them shows a full-width band of that pale
 * ground. Between a dark chapter and a light one that is correct and wanted. Between two dark
 * chapters it is a pale stripe across the page with nothing in it.
 *
 * MEASURED ON STAGING AT 1440 WIDE, not read off the rules. There were exactly two: 104px between
 * the Kogi story and "Three simple principles", and 88px between the time chapter and the
 * triptych. Nothing on the page paints #FFFFFF, so an earlier diagnosis that the light chapters had
 * lost their ground was looking at the wrong thing.
 *
 * THE DESIGN FILE SETTLES WHAT SHOULD BE THERE, and it is nothing at all. Measured in
 * EARTHCELL-02-DESIGN-REFERENCE.dc.html: every dark chapter that follows another dark chapter has
 * margin-top 0 and the two boxes touch. Space above a LIGHT chapter is padding, so it falls inside
 * that chapter and lands on its own ground. Space above a dark chapter that follows a light one
 * stays a margin, because the ground showing through it is the light chapter's and that is the
 * intended pairing.
 *
 * So the rule is only about the dark-after-dark case. The margins everywhere else are correct and
 * are deliberately left alone.
 *
 * ONE SELECTOR RATHER THAN THE THIRTY PAIRS, because the chapter order still moves. The partners
 * chapter is not built yet and it goes between two of these. A list of pairs would be right today
 * and quietly wrong the moment anything is inserted.
 */
.earthcell-content :is(.ec-c1, .ec-c3, .ec-c4, .ec-c5, .ec-c6, .ec-c8, .ec-band)
	+ :is(.ec-c1, .ec-c3, .ec-c4, .ec-c5, .ec-c6, .ec-c8, .ec-band) {
	margin-top: 0;
}

/* Prose and a caveat or a list beside it. */
.earthcell-content .ec-grid--sidebar {
	grid-template-columns: minmax(0, 1fr) 380px;
	gap: 56px;
}

/* The spatial data model diagram and the monitoring readouts beside it. */
.earthcell-content .ec-grid--readouts {
	grid-template-columns: minmax(0, 1fr) 300px;
	gap: 56px;
	margin-top: 72px;
}

/* Two things that answer each other, read side by side. */
.earthcell-content .ec-grid--pair {
	grid-template-columns: repeat(2, minmax(0, 1fr));
	gap: 64px;
}

/* --------------------------------------------------------------------------
   A heading and its body. The reading column.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-col {
	max-width: var(--ec-measure);
}

.earthcell-content .ec-col > h2 {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(28px, 4vw, 40px);
	line-height: 1.1;
	font-weight: 800;
	letter-spacing: -0.03em;
}

.earthcell-content .ec-col > p {
	margin: 0;
	max-width: none;
	font-size: 18px;
	line-height: 1.65;
}

/*
 * THE STACK GAP HAS TO BE AS SPECIFIC AS THE RULE THAT ZEROES THE MARGIN, or it loses.
 *
 * This was `.ec-col > * + *`. Two class names and two universal selectors score 0-2-0. The
 * `margin: 0` on `.ec-col > p` above it scores 0-2-1, because a tag name outranks a `*`. So every
 * paragraph after the first kept a zero margin and the whole page ran its sentences together with
 * no space between them. It looked like the stylesheet had not loaded. It had; this one rule was
 * being overruled by the one four lines above it.
 *
 * Naming the tags in the selector puts it at 0-2-2, above the rule it has to beat. Do not go back
 * to `* + *` here, and do not fix it with `!important`.
 */
.earthcell-content .ec-col > p + p,
.earthcell-content .ec-col > p + ul,
.earthcell-content .ec-col > p + ol,
.earthcell-content .ec-col > ul + p,
.earthcell-content .ec-col > ol + p,
.earthcell-content .ec-col > * + * {
	margin-top: 18px;
}

.earthcell-content .ec-col > h2 + p {
	margin-top: 24px;
}

/* --------------------------------------------------------------------------
   The introduction. Running text at the reading column.

   NO TOP PADDING, unlike the design's other bands. The design puts 88px above this block because
   in the design the map is directly above it. On the built page the theme prints the page title
   and its standfirst in between, and .earthcell-head already carries that 88px. A second one puts
   an empty screen between the standfirst and the first sentence it introduces.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-intro {
	max-width: var(--ec-measure);
}

.earthcell-content .ec-intro > p {
	margin: 20px 0 0;
	max-width: none;
	font-size: 19px;
	line-height: 1.65;
}

.earthcell-content .ec-intro > p:first-child {
	margin-top: 0;
}

.earthcell-content .ec-intro .ec-lead {
	font-size: 21px;
	line-height: 1.6;
}

/*
 * 640px IS A CEILING, NOT A WIDTH, and treating it as a width put the button underneath.
 *
 * Measured when the band was 1056px wide with 40px of padding each side, so 976px of room. The
 * sentence asked for its full 640, the button measures 362, and 640 + 40 of gap + 362 is 1042.
 * Flex had nothing to take from either side, so it wrapped and the whole right half of the lime
 * band went empty with the button sitting under the text on the left. The band is narrower still
 * since 4 September 2026, so the ceiling matters more rather than less.
 *
 * The design puts the button on the right. The sentence is the part that can give ground, so it
 * is the part allowed to shrink.
 */
.earthcell-content .ec-cta__text {
	flex: 1 1 320px;
	min-width: 0;
	max-width: 640px;
}


.earthcell-content .ec-cta__title {
	margin: 0;
	font-size: 28px;
	line-height: 1.15;
	font-weight: 800;
	letter-spacing: -0.025em;
}

.earthcell-content .ec-cta__line {
	margin: 10px 0 0;
	max-width: none;
	font-size: 17px;
	line-height: 1.55;
}

.earthcell-content .ec-cta__action {
	flex: 0 0 auto;
	margin: 0;
	max-width: none;
}

/* --------------------------------------------------------------------------
   Buttons. Two of them on the page, and they are opposite colours because they sit on opposite
   grounds: teal on the lime band, lime on the dark enquiry band.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-button {
	display: inline-block;
	padding: 16px 28px;
	border: 0;
	font-family: inherit;
	font-size: 16px;
	font-weight: 700;
	line-height: 1.2;
	text-decoration: none;
	cursor: pointer;
}

.earthcell-content .ec-button--teal {
	background: var(--ec-teal);
	color: var(--ec-paper);
}

.earthcell-content .ec-button--lime {
	background: var(--ec-lime);
	color: var(--ec-ground);
}

.earthcell-content .ec-contrast__col {
	padding-left: 20px;
	border-left: 2px solid;
}

.earthcell-content .ec-contrast__col--is {
	border-left-color: var(--ec-lime);
}

.earthcell-content .ec-contrast__col--isnot {
	border-left-color: var(--ec-aqua);
}

.earthcell-content .ec-contrast__label {
	margin: 0;
	max-width: none;
	font-size: 18px;
	font-weight: 700;
	letter-spacing: -0.01em;
}

/*
 * THE THEME DRAWS A LIME ARROW ON EVERY LIST ITEM ON THE SITE, and it is the wrong mark here.
 *
 * The rule is `.entry-content li::before { background-image: url(img/circle-arrow-right.svg) }`,
 * a 24px lime disc. It is right for a page that is selling something. On this page it appeared
 * against "an offset" and "a tradable financial asset" in the "An EarthCell is not" column, where
 * a green forward arrow reads as approval of the exact six things the column denies. It also
 * covered the plain ruled rows of the two "could return and see" lists.
 *
 * Every list inside .earthcell-content gets its own mark below. None of them get the arrow.
 */
.earthcell-content .ec-contrast li::before,
.earthcell-content .ec-returns__list li::before {
	content: none;
	display: none;
}

.earthcell-content .ec-contrast__col--is li::after {
	background: var(--ec-lime);
}

.earthcell-content .ec-contrast__col--isnot li::after {
	background: var(--ec-aqua);
}

/* --------------------------------------------------------------------------
   The two boxes that hold a limit rather than an argument.

   The reporting limits and the illustrative-history caveat. Both were bold body text, which is
   the one treatment that makes a reader skip a sentence. Both are bordered boxes now, at the
   reading column, in the pattern the design uses for the same job on its dark bands (section 0B.6).

   THE TEXT IS TEAL, NOT AQUA. Aqua is the design's demonstration colour and these boxes carry the
   demonstration marking, but #00EED2 on #F4F8F9 is unreadable. The border does the marking and
   the teal token does the reading.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-disclaimer {
	max-width: var(--ec-measure);
	margin: 40px 0 0;
	padding: 18px 20px;
	border: 1px solid rgba(0, 238, 210, 0.6);
	background: rgba(0, 238, 210, 0.06);
	font-size: 16px;
	line-height: 1.6;
	color: var(--ec-teal);
}

.earthcell-content .ec-band__inner {
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: 88px var(--ec-gutter);
}

.earthcell-content .ec-band--teal {
	background: var(--ec-teal);
	color: var(--ec-paper);
}

.earthcell-content .ec-band--dark {
	background: var(--ec-ground);
	color: #e6eef0;
}

.earthcell-content .ec-band--teal h2,
.earthcell-content .ec-band--dark h2 {
	color: var(--ec-paper);
}

.earthcell-content .ec-band--teal .ec-col > p {
	color: rgba(244, 248, 249, 0.92);
}

.earthcell-content .ec-band--dark .ec-col > p {
	color: rgba(230, 238, 240, 0.82);
}

/* The enquiry heading is the largest on the page, because it is the page's ask (section 0B.8). */
.earthcell-content .ec-band--dark .ec-col > h2 {
	font-size: clamp(30px, 4.4vw, 44px);
}

/* --------------------------------------------------------------------------
   The imagery credit, wherever a chapter map needs one.

   SHOWING IT IS THE CONDITION ESRI'S IMAGERY IS USED ON, so it is not a preference and no rule
   below may hide it, fade it or position anything over it.

   IT IS A LINE OF THE PAGE, NOT A CONTROL OF THE MAP. Every chapter map is built with the MapLibre
   attribution control turned off, because that control sits in the bottom right corner of its own
   canvas, and the site's chat button is fixed to the bottom right corner of the window: 60px
   square, 35px in and 20px up. It has covered the credit twice already.

   TWELVE PIXELS, NOT TEN, AND A LIGHTER GREY. The design's own floor for a caption is 10px and
   #5E6C74 was inside it, but measured against the near-black ground that pair comes out at about
   3.6 to 1, and the minimum for text is 4.5. The design review of 29 August 2026 counted six
   instances of it. 12px on #A8B6BE clears the minimum and the line still reads as a caption.

   THE 96px OF RIGHT PADDING IS THAT CHAT BUTTON, AND 64px WAS NOT ENOUGH.

   The button is 60px wide and it sits 35px in from the right edge on a phone and 20px in on a
   desktop, so it occupies the rightmost 95px of the window at its worst. 64px cleared it on a
   desktop, where the reading column stops well short of the window edge anyway, and failed on a
   390px phone, where the column runs almost to the edge: five glyphs of the C6 credit and two of
   the C5 caption were behind it. Measured with document.elementFromPoint across the glyphs of
   each line, using a Range over the text node, on staging on 29 August 2026. Not by looking.

   96px is the button's own 95px rounded up. It costs nothing at desktop, where these lines are
   short and the column ends 120px clear of the button already.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-credit {
	margin: 0;
	max-width: none;
	padding-right: 96px;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1.6;
	letter-spacing: 0.12em;
	color: #a8b6be;
}

/* ==========================================================================
   C3 RELATIONSHIP. A live map running to the edge of its own column.

   WHAT THIS REPLACES. Prose in a light band with a bordered caveat beside it, which is the
   superseded article design. Round 3 makes it a full-bleed near-black chapter whose left column is
   a map with no border and no padding, touching the top, bottom and left edges of the band's
   inner block. Piers reported the page as having none of the design's full-bleed pictures. This is
   the second of the three.

   THE MAP HAS NO BORDER AND THE COLUMN HAS NO GAP AT ITS EDGE. That is what "running to the edge
   of its own column" means, and it is what makes the chapter read as one picture with words beside
   it rather than as a picture in a box. The vertical padding is on the words, not on the map, so
   the map is as tall as the section.

   THE COLUMNS COLLAPSE ON THEIR OWN. minmax(min(100%, 340px), 1fr) puts them side by side while
   both fit and stacks them when they stop fitting. There is no breakpoint here and there must not
   be one.
   ========================================================================== */

.earthcell-content .ec-c3 {
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: 104px;
	background: var(--ec-ground);
	color: #e6eef0;
}

.earthcell-content .ec-c3__inner {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 340px), 1fr));
	gap: clamp(24px, 4vw, 56px);
	align-items: stretch;
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: 0 var(--ec-gutter);
}

/*
 * THE .earthcell-content PREFIX IS LOAD-BEARING, for the same reason it is on C1's map div.
 * MapLibre adds .maplibregl-map to this element and its own stylesheet sets position on it at a
 * specificity of 0-1-0. This rule scores 0-2-0 and wins outright.
 *
 * THE MINIMUM HEIGHT IS WHAT GIVES THE MAP A CANVAS AT ALL. A grid cell with nothing in it has no
 * height, and a MapLibre container with no height renders a canvas of nothing and reports no
 * error. When the two columns stack, this minimum is the whole height of the map.
 */
.earthcell-content .ec-c3__map {
	position: relative;
	min-height: clamp(300px, 46vh, 560px);
	background: var(--ec-ground);
}

.earthcell-content .ec-c3__words {
	display: flex;
	flex-direction: column;
	gap: 26px;
	padding: clamp(40px, 7vw, 88px) 0;
}

.earthcell-content .ec-c3__title {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(26px, 12.3px + 2.63vw, 46px);
	line-height: 1.06;
	font-weight: 800;
	letter-spacing: -0.03em;
	color: var(--ec-paper);
}

.earthcell-content .ec-c3__body {
	margin: 0;
	/*
	 * IT WAS UNCAPPED. In two columns the grid held it near the reading measure anyway. In one
	 * column, which is everything below about 940px, it ran the full width of the band and came out
	 * at 819px, about 75 characters a line. The cap is the page's one reading column now; it was a
	 * chapter-local 620px until 4 September 2026.
	 */
	max-width: var(--ec-measure);
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
	color: rgba(230, 238, 240, 0.82);
}

/* ==========================================================================
   THE THREE SITE CARDS.

   THIS SECTION USED TO BE THE EMPTIEST PART OF THE PAGE. Each site put its name in one column and
   its body in another, leaving about 350 pixels of nothing beside every one of them, in the one
   place a funder decides whether they care. Piers reported it and a design review counted the
   space. It is three cards with a diagram each now.

   THE DIAGRAMS ARE NOT PHOTOGRAPHS AND NOT MAPS. There is no photograph of any of the three sites
   and no boundary for two of them. They are the page's own hexagons at three relative sizes, and
   the line under the row says in words that they are not site boundaries. That line is the reason
   they are allowed to exist.

   THREE UP, THEN TWO, THEN ONE, WITH NO BREAKPOINT. minmax(min(100%, 280px), 1fr) puts as many in
   a row as can hold 280px each. A card is never orphaned two-up-one because the diagram keeps every
   card the same height.

   280px AND NOT 260px, AND THE 20 PIXELS MATTER. At 260 the arithmetic just allowed three columns
   at a 900px window: the band is 876px wide there and three 260px cards with two 28px gaps come to
   876 exactly. So the whole 760 to 940 range got three cramped cards rather than two comfortable
   ones, which is the fault the design review raises about that band. At 280 the same sum comes to
   896, three no longer fit, and the flip to three-up happens at about 975px instead.
   ========================================================================== */

.earthcell-content .ec-sitecards {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 280px), 1fr));
	gap: clamp(20px, 3vw, 32px);
	margin-top: 40px;
}

.earthcell-content .ec-sitecard {
	display: flex;
	flex-direction: column;
	gap: 14px;
	max-width: none;
}

/*
 * FOUR BY THREE, AND THE HONEYCOMB FILLS IT.
 *
 * preserveAspectRatio on the SVG is set to slice, so the lattice is cropped by the frame rather
 * than letterboxed inside it. A honeycomb with a visible edge reads as an object; one that runs
 * off all four sides reads as ground continuing past the picture, which is what it is.
 */
.earthcell-content .ec-sitecard__diagram {
	position: relative;
	aspect-ratio: 4 / 3;
	overflow: hidden;
	background: var(--ec-ground);
}

/*
 * REAL GROUND UNDER THE LATTICE. Added 30 August 2026, and it is the point of the whole card.
 *
 * THREE TREATMENTS WERE TRIED AND FAILED BEFORE THIS ONE. Black behind the hexagons gave the cards
 * no sense of place at all. A photograph washed in at 34 per cent gave what the client called a
 * murky black image with some cells on it. A six by six grid of clipped spans produced hexagons 55
 * to 100 pixels across that read as a hexagonal cut-out in a field rather than as cells.
 *
 * What makes these read as measurement rather than decoration is real imagery with a fine lattice
 * over it. DO NOT COARSEN THE HONEYCOMB to make it more visible: coarse is the failure mode.
 *
 * FOUR STATIC TILES IN A SQUARE, NOT A LIVE MAP. Three small maps on a page that already carries
 * six of them is six more WebGL contexts for three pictures that never move. These are ordinary
 * images, lazily loaded, and they cost nothing until the row is scrolled to.
 *
 * ONE PIECE OF GROUND AT THREE ZOOM LEVELS, not three places: z16 for Munekan Masha, z15 for
 * Duanama, z14 for Esmeralda, each 2x2 tiles around the same point. The card with the largest
 * region gets the widest view. Every tile was fetched and checked before being written in.
 *
 * PINNED THROUGH THE MIDDLE. The mosaic is square and the frame is 4:3, so it is taller than the
 * frame. Centring it vertically crops evenly top and bottom instead of dropping the bottom third.
 */
.earthcell-content .ec-sitecard__tiles {
	position: absolute;
	top: 50%;
	left: 0;
	width: 100%;
	transform: translateY(-50%);
	display: grid;
	grid-template-columns: 1fr 1fr;
	line-height: 0;
	filter: brightness(1.14) saturate(1.06) contrast(1.03);
}

.earthcell-content .ec-sitecard__tiles img {
	display: block;
	width: 100%;
	height: auto;
	max-width: none;
}

/*
 * A SCRIM OVER THE WHOLE PICTURE, not only its foot. The lime lattice has to stay readable across
 * all of it, and satellite imagery of forest at midday is bright enough in places to lose a .7px
 * stroke. It is light at the top and heavier at the bottom, where the scale label sits.
 */
.earthcell-content .ec-sitecard__diagram::after {
	content: "";
	position: absolute;
	inset: 0;
	pointer-events: none;
	background: linear-gradient(180deg, rgba(5, 8, 11, 0.34), rgba(5, 8, 11, 0.62));
}

/*
 * THE LATTICE SITS OVER THE IMAGERY AND OVER THAT SCRIM, so the cells read as drawn on the ground
 * rather than buried under a wash. Without the z-index it would be painted first and the scrim
 * would dim the one thing the card exists to show.
 */
.earthcell-content .ec-sitecard__hex {
	position: absolute;
	inset: 0;
	z-index: 1;
	display: block;
	width: 100%;
	height: 100%;
}

/*
 * THE SCRIM IS ONLY AT THE FOOT, where the scale label sits. Over the whole diagram it would dim
 * the thing the card exists to show.
 */
/*
 * THE z-index IS NOT DECORATION. The lattice above is raised so it draws over the imagery, and a
 * positioned element with z-index beats a positioned sibling without one whatever the source order
 * says. Without these two numbers the foot scrim and the scale label would be painted underneath a
 * honeycomb, and the label is the line that says these are scales and not boundaries.
 */
.earthcell-content .ec-sitecard__scrim {
	position: absolute;
	left: 0;
	right: 0;
	bottom: 0;
	z-index: 2;
	height: 45%;
	pointer-events: none;
	background: linear-gradient(180deg, rgba(5, 8, 11, 0) 0%, rgba(5, 8, 11, 0.9) 100%);
}

.earthcell-content .ec-sitecard__scale {
	position: absolute;
	left: clamp(12px, 2vw, 18px);
	right: clamp(12px, 2vw, 18px);
	bottom: clamp(10px, 1.6vw, 14px);
	z-index: 3;
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.5;
	letter-spacing: 0.12em;
	color: #8c9ba3;
}

.earthcell-content .ec-sitecard__name {
	margin: 0;
	font-size: clamp(22px, 4.4vw, 28px);
	line-height: 1.15;
	font-weight: 800;
	letter-spacing: -0.02em;
}

.earthcell-content .ec-sitecard__body {
	margin: 0;
	max-width: none;
	font-size: clamp(15px, 2.1vw, 16px);
	line-height: 1.6;
}

/*
 * THE LINE THAT MAKES THE DIAGRAMS HONEST. It is not decoration and it is not a caption. Without
 * it three pictures of hexagons under three site names read as three site boundaries, which is
 * exactly what they are not. If real boundaries ever arrive, the diagrams become real maps and
 * this line comes off with them.
 */
.earthcell-content .ec-sitecards__note {
	margin: 22px 0 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.12em;
	color: #5e6c74;
}

/* ==========================================================================
   C4 KOGI STORY. Full-bleed teal, and the photograph fills the whole section.

   WHAT THIS REPLACES, AND WHY IT WAS WRONG. Until 29 August 2026 the photograph sat in a
   four-by-three slot in the right column of a two-column teal band. That is the article layout,
   and IMPLEMENTATION-GAPS.md says in its own opening section that the article design no longer
   exists. Piers reported the fault as the page having none of the design's full-bleed background
   pictures, and he was right: this was the first of them.

   THE GRADIENT IS DOING THE WORK, NOT A DARKER PHOTOGRAPH. It runs at 105 degrees, nearly opaque
   on the left, still 86 per cent at the halfway point, and thin by the right edge. So the left
   two thirds are a dark teal ground that body text sits on comfortably, and the right third is a
   photograph. One picture does both jobs.

   NO BREAKPOINTS. The type is clamp(), and the locator row wraps on flex when the locator and its
   notes stop fitting side by side. Below about 700px the notes drop under the locator on their
   own, which is the design's own behaviour rather than a phone rule.

   THE PHOTOGRAPH IS CREDITED IN THE PAGE, NOT IN A FIGCAPTION. A background image has nothing to
   put a caption under, so the credit is the first of the two monospace notes beside the locator.
   The Esri credit for the locator tiles is the second, and neither may be removed: showing them
   is the condition both pictures are used on.
   ========================================================================== */

/*
 * FULL BLEED WITH THE SAME PLAIN 100vw SUM THE BANDS USE, and for the same reason set out above
 * them: --ec-scrollbar carries tokens rather than a measured width, and inside the reading column
 * the sum cancels itself out and the section stays as wide as the prose.
 *
 * overflow: hidden IS SAFE HERE, unlike on C1. Nothing in this section is sticky. It is what keeps
 * the photograph inside the section when object-fit crops it.
 */
/*
 * C4: THE SAME GROUND AS THE INSTRUMENT, WITH THE PHOTOGRAPH AS A PANEL BESIDE THE PROSE.
 *
 * REBUILT 30 AUGUST 2026, AND THE OLD SHAPE HAD TWO FAULTS.
 *
 * It was teal. That made it the only saturated colour block on the page, appearing once, directly
 * under the black instrument, and it read as a blue slab across the window.
 *
 * And the photograph ran to all four edges under a 105-degree gradient that darkened the left so
 * type could sit on it. The map above this section also runs to both edges, so once the pale band
 * between them was closed the top of the page was two edge-to-edge pictures meeting at a seam,
 * which reads as one broken image. Piers sent a screenshot of it the same morning.
 *
 * The ground is now the instrument's own #05080B and a hairline marks the change from instrument to
 * document. Nothing is layered: the words hold the left, the picture is a figure on the right at its
 * own size, and neither is on top of the other.
 *
 * THE min-height IS GONE. It existed to give a background image something to fill. A section sized
 * by its own content cannot have a hole in it.
 */
.earthcell-content .ec-c4 {
	position: relative;
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: 88px;
	overflow: hidden;
	background: var(--ec-ground);
	color: #e6eef0;
	border-top: 1px solid rgba(230, 238, 240, 0.12);
}

/*
 * The teal behind it is not decoration either. A photograph that has not arrived, or a reader who
 * has images switched off, leaves the section on the project's own colour with white text on it,
 * which reads. A transparent section would leave white text on white.
 */
/*
 * A PICTURE AT ITS OWN SIZE, NOT A BACKGROUND FILLING A BOX.
 *
 * "contain" rather than "cover", because a 4:3 photograph forced into a taller box loses its sides,
 * and the sides of this one are where the people standing at the boulder are.
 */
.earthcell-content .ec-c4__photo {
	display: block;
	width: 100%;
	height: auto;
	max-width: 100%;
	object-fit: contain;
}

/*
 * THE READABILITY GRADIENT IS GONE, and the element it painted no longer exists in the markup. This
 * rule is kept as a tombstone for one release: if a saved copy of the old page body is ever pushed,
 * the wash div comes back with it, and without this it would paint a translucent slab over the whole
 * chapter. Delete it once the page body has been stable for a while.
 */
.earthcell-content .ec-c4__wash {
	display: none;
}

/*
 * ASYMMETRIC, PROSE WIDER THAN THE PICTURE. At 1440 that is about 587px of words against a 405px
 * photograph. The words are the chapter; the photograph is evidence beside them.
 *
 * ONE COLUMN BELOW 900px. Stacked, the reading order is words then picture then caption, which is
 * the order somebody reads them in anyway.
 */
.earthcell-content .ec-c4__inner {
	position: relative;
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	gap: clamp(30px, 5vw, 64px);
	align-items: start;
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: clamp(60px, 10vw, 104px) var(--ec-gutter);
}

@media (min-width: 900px) {
	.earthcell-content .ec-c4__inner {
		grid-template-columns: minmax(0, 1.45fr) minmax(0, 1fr);
	}
}

.earthcell-content .ec-c4__words {
	display: flex;
	flex-direction: column;
	gap: 26px;
}

.earthcell-content .ec-c4__figure {
	display: flex;
	flex-direction: column;
	gap: 12px;
	margin: 0;
	max-width: 405px;
}

.earthcell-content .ec-c4__title {
	margin: 0;
	max-width: var(--ec-heading);
	/*
	 * THE MIDDLE TERM LOOKS ODD AND IT IS DELIBERATE. It was 6.8vw, which reaches this clamp's
	 * 52px ceiling at a 764px window, so every window from a small laptop upwards got the largest
	 * heading on the page. At 900px that is 52px type over an 819px measure, which is about 75
	 * characters a line and reads as a phone layout at desktop sizes. The design review of
	 * 29 August 2026 calls it the one remaining structural gap.
	 *
	 * 11.6px + 3.16vw gives 36px at 760, 40px at 900 and the full 52px by 1280, which is the step
	 * the review asks for. It still bottoms out at 28px on a phone.
	 *
	 * IT IS A CLAMP AND NOT A BREAKPOINT ON PURPOSE. Round 4 of the handoff says this band is
	 * designed by re-cutting the clamps, and a breakpoint here would put a visible jump in the
	 * middle of the range it is meant to smooth.
	 */
	font-size: clamp(28px, 11.6px + 3.16vw, 52px);
	line-height: 1.04;
	font-weight: 800;
	letter-spacing: -0.03em;
	color: var(--ec-paper);
}

.earthcell-content .ec-c4__body {
	margin: 0;
	max-width: var(--ec-measure);
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
	color: rgba(244, 248, 249, 0.92);
}

.earthcell-content .ec-c4__row {
	display: flex;
	flex-wrap: wrap;
	align-items: flex-end;
	gap: 20px;
	margin-top: 8px;
}

/* --------------------------------------------------------------------------
   The satellite locator. Four still pictures in a frame, and not a map.

   IT IS NOT A MapLibre INSTANCE AND MUST NOT BECOME ONE. A sixth live map on this page would cost
   a WebGL context and a tile budget to say what four still images say for nothing: that the story
   happens at a real, findable place.

   TWO THINGS IT MUST NOT BECOME, both of them review defects already made once. A single tile at
   zoom 11 is mostly shadowed mountain and reads as a black square. Dimming it with opacity gives
   the same result. So: four tiles at zoom 13, brightened, at full opacity.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c4__locator {
	display: flex;
	flex-direction: column;
	border: 1px solid rgba(0, 238, 210, 0.45);
}

.earthcell-content .ec-c4__frame {
	position: relative;
	width: 212px;
	height: 212px;
	overflow: hidden;
	background: #04262b;
}

/*
 * THE BRIGHTENING IS ON THE MOSAIC AND NOT ON THE FRAME. On the frame it would also brighten the
 * aqua ring and the marker dot, which are meant to read as instrument marks over the picture
 * rather than as part of it.
 */
.earthcell-content .ec-c4__mosaic {
	display: grid;
	grid-template-columns: repeat(2, 106px);
	grid-template-rows: repeat(2, 106px);
	filter: brightness(1.18) saturate(1.12) contrast(1.06);
}

.earthcell-content .ec-c4__tile {
	display: block;
	width: 106px;
	height: 106px;
	max-width: none;
}

/*
 * WORDPRESS TURNS A SINGLE NEWLINE INSIDE A BLOCK INTO A <br>, AND INSIDE A GRID EVERY <br> TAKES
 * A CELL OF ITS OWN.
 *
 * The four tiles were written one per line, so wpautop put three line breaks between them. The
 * grid then held seven items instead of four, laid them out two across and four down, and the
 * frame's overflow: hidden showed the top half of a column of tiles. On screen that reads as a
 * locator half the width it should be. Found on staging on 29 August 2026, not by reading the
 * markup.
 *
 * THE REAL FIX IS IN THE PAGE, where the four tags now sit on one line, and this rule is the
 * guard. It is the same family of fault as the p:empty rule further down this file: WordPress
 * adds a node the markup never asked for, and a grid gives it a cell.
 */
.earthcell-content .ec-c4__mosaic br,
.earthcell-content .ec-c5__years br {
	display: none;
}

/*
 * AND A LINE BREAK DIRECTLY INSIDE A CHAPTER SECTION, WHICH IS A DIFFERENT FAULT WITH THE SAME
 * CAUSE.
 *
 * wpautop also leaves a line break after an HTML comment. In C4 that put one line box above the
 * content block, on top of the padding the design gives it, so the chapter began lower than the
 * others and nothing in the markup said why. Measured on staging on 29 August 2026.
 *
 * None of these sections ever wants a line break as a direct child. Every one of them holds
 * positioned layers and one content block, and a stray line box between them is always the page
 * builder's and never the design's. The only legitimate break in this whole chapter is the one
 * inside the locator's coordinates, which is two lines on purpose and is not a direct child of a
 * section.
 */
.earthcell-content .ec-c1 > br,
.earthcell-content .ec-c3 > br,
.earthcell-content .ec-c4 > br,
.earthcell-content .ec-c5 > br,
.earthcell-content .ec-c6 > br,
.earthcell-content .ec-c10 > br {
	display: none;
}

.earthcell-content .ec-c4__ring {
	position: absolute;
	left: 32%;
	top: 45%;
	width: 26px;
	height: 26px;
	margin: -13px 0 0 -13px;
	border: 1.5px solid var(--ec-aqua);
	border-radius: 50%;
	box-shadow: 0 0 0 1px rgba(0, 24, 29, 0.55);
}

.earthcell-content .ec-c4__dot {
	position: absolute;
	left: 32%;
	top: 45%;
	width: 4px;
	height: 4px;
	margin: -2px 0 0 -2px;
	background: var(--ec-aqua);
	border-radius: 50%;
}

.earthcell-content .ec-c4__coords {
	margin: 0;
	padding: 8px 10px;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.5;
	letter-spacing: 0.1em;
	color: var(--ec-aqua);
	background: rgba(0, 24, 29, 0.75);
}

/*
 * THE COLUMN IS CAPPED SO ITS RIGHT EDGE NEVER REACHES THE CHAT BUTTON.
 *
 * These two are bordered boxes, so 96px of padding inside them would leave a visible hole rather
 * than move the text. The whole column moves instead. On a desktop the row is wide and the cap
 * never bites, so it stays at the design's 420px. On a 390px phone the row is about 364px and the
 * cap brings the boxes in to 268px, which ends 14px clear of the button. One of the two is the
 * Esri credit for the locator tiles, so this is not a nicety.
 */
.earthcell-content .ec-c4__notes {
	display: flex;
	flex-direction: column;
	gap: 10px;
	max-width: min(420px, calc(100% - 96px));
}

/*
 * THE DESIGN'S TWO SQUARE-BRACKET PLACEHOLDERS DO NOT RENDER, on Piers's instruction of 28 August
 * 2026, twice. What renders in the same boxes is the two credits, which are not placeholders and
 * are not optional: one names the photograph, the other is Esri's condition for the tiles.
 */
/*
 * THE PHOTOGRAPH'S CREDIT IS A CAPTION NOW, NOT A BOX ON TOP OF A PICTURE.
 *
 * It used to carry a frame and a dark wash because it floated over a full-bleed photograph and had
 * to hold itself apart from whatever was behind it. It sits under a contained figure now, on the
 * chapter's own ground, so the box is noise: a caption directly beneath its picture is already
 * attached to it.
 *
 * #8C9BA3 IS THE PAGE'S CAPTION GREY ON DARK and it clears the contrast standard at this size. The
 * other neutral in this file, #5E6C74, is the light-ground caption colour and fails on dark.
 *
 * BOTH CLASSES ARE NAMED BECAUSE THE BASE RULE COMES AFTER THIS ONE IN THIS FILE. At equal
 * specificity the later rule wins, so a single-class modifier here is silently overruled by the
 * aqua and the border it is trying to remove. Measured on staging: the caption was still aqua and
 * still framed after this rule was deployed. This file puts a modifier before its base in more
 * than one place, so check the order before writing one.
 */
.earthcell-content .ec-c4__note.ec-c4__note--photo {
	padding: 0;
	border: 0;
	background: none;
	color: #8c9ba3;
	letter-spacing: 0.06em;
}

.earthcell-content .ec-c4__note {
	margin: 0;
	padding: 12px 16px;
	border: 1px solid rgba(0, 238, 210, 0.5);
	font-family: var(--ec-mono);
	font-size: 11px;
	line-height: 1.6;
	letter-spacing: 0.04em;
	color: var(--ec-aqua);
}

/* ==========================================================================
   C5 TIME. Come back to the same place.

   THE PICTURE DOES NOT CHANGE WHEN THE YEAR CHANGES, AND THE CAPTION SAYS SO. There is no second
   capture of this ground and we do not have one to show. What the year buttons change is which
   sample observation points are drawn on the map and which rows in the list are at full strength.
   Every rule below that reacts to a year reacts to the same data-ec-stage attribute the widget
   writes, so the map, the buttons, the rows and the caption can never disagree.

   THE STATE WITH NO ATTRIBUTE IS THE STATE A READER WITH NO SCRIPT GETS: 2026 lit, every row at
   full strength, the first caption showing. That is why every selector below names both the stage
   it wants and, for stage zero, the no-attribute case beside it.

   ROWS FOR LATER YEARS FADE RATHER THAN VANISH. A reader on 2026 can see what is coming instead of
   finding an empty list, which is the design's own reason and a good one.
   ========================================================================== */

.earthcell-content .ec-c5 {
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: 88px;
	background: var(--ec-ground);
	color: #e6eef0;
}

.earthcell-content .ec-c5__inner {
	display: flex;
	flex-direction: column;
	gap: 26px;
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: clamp(56px, 9vw, 96px) var(--ec-gutter);
}

/*
 * A COLUMN, NOT A space-between ROW, and this is the fault the spec warns about by name.
 *
 * The chip and the heading shared a row with the space pushed between them. That meant the heading
 * measured whatever the chip left it, about 518px, rather than the width every other chapter
 * heading gets. So the largest heading in the chapter wrapped early, and nothing in the markup said
 * why.
 *
 * The chip is an eyebrow above the heading now. That is also the right order to read them in: the
 * chip says this record is a demonstration, and a warning after the thing it qualifies is late.
 */
.earthcell-content .ec-c5__head {
	display: flex;
	flex-direction: column;
	align-items: flex-start;
	gap: 16px;
}

.earthcell-content .ec-c5__title {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(28px, 11.6px + 3.16vw, 52px);
	line-height: 1.04;
	font-weight: 800;
	letter-spacing: -0.03em;
	color: var(--ec-paper);
}

/*
 * THE CHIP IS NOT DECORATION. Every figure and every dot in this chapter comes from the
 * demonstration boundary, and CLAUDE.md requires that to be marked on screen while it is true.
 * It is the first of three places this chapter says so; the row statuses and the caption are the
 * other two.
 */
.earthcell-content .ec-c5__chip {
	margin: 0;
	max-width: none;
	padding: 9px 14px;
	border: 1px solid rgba(0, 238, 210, 0.45);
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.14em;
	color: var(--ec-aqua);
}

.earthcell-content .ec-c5__body {
	margin: 0;
	/* The reading measure, for the reason given at .ec-c3__body. It was 820px, then 620px. */
	max-width: var(--ec-measure);
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content .ec-c5__grid {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 290px), 1fr));
	gap: clamp(28px, 4vw, 40px);
	align-items: stretch;
	margin-top: 10px;
}

/* The prefix beats maplibre-gl.css, and the minimum height is what gives the canvas a size. */
.earthcell-content .ec-c5__map {
	position: relative;
	min-height: clamp(280px, 42vh, 440px);
	border: 1px solid rgba(230, 238, 240, 0.16);
	background: var(--ec-ground);
}

.earthcell-content .ec-c5__record {
	display: flex;
	flex-direction: column;
	gap: 18px;
}

.earthcell-content .ec-c5__years {
	display: flex;
	gap: 10px;
}

.earthcell-content .ec-c5__year {
	flex: 1;
	padding: 14px 0;
	background: transparent;
	border: 1px solid rgba(230, 238, 240, 0.2);
	border-radius: 0;
	font-family: var(--ec-mono);
	font-size: 13px;
	letter-spacing: 0.1em;
	color: rgba(230, 238, 240, 0.72);
	cursor: pointer;
}

/*
 * THE CHOSEN YEAR IS FILLED. The no-attribute case is listed beside stage zero on purpose: a
 * reader whose script never loaded should see 2026 chosen rather than three identical buttons.
 */
.earthcell-content .ec-c5:not([data-ec-stage]) .ec-c5__year[data-ec-year-button="0"],
.earthcell-content .ec-c5[data-ec-stage="0"] .ec-c5__year[data-ec-year-button="0"],
.earthcell-content .ec-c5[data-ec-stage="1"] .ec-c5__year[data-ec-year-button="1"],
.earthcell-content .ec-c5[data-ec-stage="2"] .ec-c5__year[data-ec-year-button="2"] {
	background: var(--ec-lime);
	border-color: var(--ec-lime);
	color: var(--ec-ground);
}

.earthcell-content .ec-c5__entries {
	display: flex;
	flex-direction: column;
}

.earthcell-content .ec-c5__entry {
	display: flex;
	flex-direction: column;
	gap: 6px;
	padding: 15px 0;
	border-bottom: 1px solid rgba(230, 238, 240, 0.12);
	transition: opacity 0.4s ease;
}

/* A row belonging to a year the reader has not reached yet. Faded, never hidden. */
.earthcell-content .ec-c5[data-ec-stage="0"] .ec-c5__entry[data-ec-entry-year="1"],
.earthcell-content .ec-c5[data-ec-stage="0"] .ec-c5__entry[data-ec-entry-year="2"],
.earthcell-content .ec-c5[data-ec-stage="1"] .ec-c5__entry[data-ec-entry-year="2"] {
	opacity: 0.34;
}

.earthcell-content .ec-c5__kind {
	display: flex;
	align-items: center;
	justify-content: space-between;
	gap: 12px;
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.16em;
	color: var(--ec-lime);
}

.earthcell-content .ec-c5__status {
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.12em;
	color: #8c9ba3;
}

/*
 * ONE ROW OF THE FOUR IS A REAL EVENT and it is the only one in aqua. The EarthCells really were
 * created from the project boundary. The drone flight, the field photographs and the restoration
 * activity have not happened, and each of those three says DEMONSTRATION in the grey.
 */
.earthcell-content .ec-c5__status--real {
	color: var(--ec-aqua);
}

.earthcell-content .ec-c5__what {
	margin: 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.4;
	color: rgba(230, 238, 240, 0.86);
}

.earthcell-content .ec-c5__caption {
	margin: 0;
	max-width: none;
	/* The chat button again. See .ec-credit above for where the 96px comes from. */
	padding-right: 96px;
	font-family: var(--ec-mono);
	font-size: 12px;
	/*
	 * THE SECOND CAPTION IS BIGGER THAN THE FIRST, AND THAT IS THE POINT.
	 *
	 * It is the sentence that says the imagery does not change between the years, on a control
	 * whose whole job is to change the year. A reader who presses 2028, sees the same picture and
	 * does not find that sentence concludes the page is broken. The design review asks for it at 13
	 * rather than 10; it is 12 here already for contrast, and the later one goes to 13 so it reads
	 * as a statement rather than as a credit.
	 */
	line-height: 1.6;
	letter-spacing: 0.1em;
	color: #a8b6be;
}

/*
 * TWO CAPTIONS, BOTH IN THE PAGE, ONE SHOWING AT A TIME.
 *
 * Both sentences are in the HTML so a reader with no script, and a search engine, get the honest
 * one for the year that is shown. Writing the text from JavaScript instead would take the second
 * sentence off the page entirely, and the second sentence is the one that says no new capture
 * exists.
 */
.earthcell-content .ec-c5__caption--later {
	display: none;
	font-size: 13px;
	color: #b9c6cc;
}

.earthcell-content .ec-c5[data-ec-stage="1"] .ec-c5__caption--first,
.earthcell-content .ec-c5[data-ec-stage="2"] .ec-c5__caption--first {
	display: none;
}

.earthcell-content .ec-c5[data-ec-stage="1"] .ec-c5__caption--later,
.earthcell-content .ec-c5[data-ec-stage="2"] .ec-c5__caption--later {
	display: block;
}

/*
 * PIERS'S LIST OF WHAT AN EARTHCELL COULD ACCUMULATE, on the dark ground of this chapter.
 *
 * The shared .ec-marker and .ec-returns__list rules were written for the light page: teal on
 * paper, and a dividing rule in the ink colour. Both disappear on near-black, so this chapter
 * restates the three colours and nothing else about them.
 */
/*
 * THE ONE LINE THE WHOLE EVIDENCE ARGUMENT NOW RESTS ON.
 *
 * Piers wrote it on 29 August 2026 and it replaced two closing paragraphs, a nine-item list and a
 * whole section explaining digital MRV twice. It is set as a statement rather than as body copy
 * because it is the conclusion of the chapter, not part of its explanation.
 */
/*
 * IT CLOSES THE CHAPTER, SO IT IS RULED OFF FROM IT. Corrected 30 August 2026.
 *
 * It was lime at 820px with no rule, which made it read as a fourth caption in a chapter that is
 * already full of small aqua and lime notes. It is the conclusion, not another annotation.
 *
 * THE RULE IS ON THE TOP EDGE, NOT THE LEFT. Both page callouts were turned that way on 29 August
 * so their text keeps a single left edge with everything above it. A left rule on a full-measure
 * block indents the whole paragraph away from the column it belongs to.
 *
 * NEAR-WHITE, NOT LIME. Lime is this page's accent for controls and for the one word in a label
 * that has to be found. A whole sentence in it competes with the enquiry button further down, and
 * the lime rule above the line already carries the emphasis.
 */
.earthcell-content .ec-c5__strapline {
	margin: 0;
	max-width: var(--ec-measure);
	padding-top: clamp(20px, 3vw, 26px);
	border-top: 2px solid var(--ec-lime);
	font-size: clamp(19px, calc(12px + 0.95vw), 27px);
	line-height: 1.35;
	font-weight: 600;
	letter-spacing: -0.02em;
	color: #f4f8f9;
}

.earthcell-content .ec-c5__returns {
	max-width: var(--ec-measure);
	margin-top: 10px;
}

.earthcell-content .ec-c5__returns .ec-marker {
	color: var(--ec-lime);
}

.earthcell-content .ec-c5__returns .ec-returns__list li {
	border-bottom-color: rgba(230, 238, 240, 0.12);
	color: rgba(230, 238, 240, 0.86);
}

/* ==========================================================================
   C6 TRIPTYCH. Three full-bleed panels, each its own map at its own scale.

   THREE REAL ZOOM SCALES OF ONE PLACE, and that is the entire chapter: 12.4, 15.2 and 16.9 on the
   demonstration cell. It is not three photographs and it must not become three photographs. The
   satellite panel shows imagery that exists today. The drone panel shows the scale a repeat flight
   would give, and nobody has flown one. The ground panel shows the scale a person standing there
   would see, and no field evidence has been recorded. The two honesty lines under those panels are
   why the chapter is allowed to show them at all.

   THE PANELS STACK ON THEIR OWN. The flex basis is min(100%, 300px), so three sit in a row while
   each can hold 300px and become three full-width panels when they cannot. There is no breakpoint.

   HOVER IS A DESKTOP NICETY AND NOTHING DEPENDS ON IT. A panel under the pointer grows and the
   other two give up the room. On a touch screen the pointer never arrives and the three stay
   equal, which is the same chapter.
   ========================================================================== */

.earthcell-content .ec-c6 {
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: 88px;
	background: var(--ec-ground);
	color: #e6eef0;
}

.earthcell-content .ec-c6__row {
	display: flex;
	flex-wrap: wrap;
	gap: 2px;
}

.earthcell-content .ec-c6__panel {
	position: relative;
	flex: 1 1 min(100%, 300px);
	min-height: clamp(300px, 36vh, 480px);
	overflow: hidden;
	transition: flex-grow 0.5s ease;
}

.earthcell-content .ec-c6__panel:hover {
	flex-grow: 1.55;
}

/* The prefix beats maplibre-gl.css. The ground colour is what shows before the first tile. */
.earthcell-content .ec-c6__map {
	position: absolute;
	inset: 0;
	z-index: 0;
	background: var(--ec-ground);
}

/*
 * DARK AT THE FOOT, WHERE THE WORDS ARE, AND CLEAR ACROSS THE MIDDLE, where the picture is. The
 * ground panel carries a heavier one because at zoom 16.9 the imagery is at its brightest and its
 * busiest, and the caption was hard to read over it.
 */
.earthcell-content .ec-c6__scrim {
	position: absolute;
	inset: 0;
	z-index: 1;
	pointer-events: none;
	background: linear-gradient(180deg, rgba(5, 8, 11, 0.35) 0%, rgba(5, 8, 11, 0) 40%, rgba(5, 8, 11, 0.9) 100%);
}

.earthcell-content .ec-c6__scrim--deep {
	background: linear-gradient(180deg, rgba(5, 8, 11, 0.4) 0%, rgba(5, 8, 11, 0.15) 40%, rgba(5, 8, 11, 0.92) 100%);
}

.earthcell-content .ec-c6__caption {
	position: absolute;
	left: 0;
	right: 0;
	bottom: 0;
	z-index: 2;
	display: flex;
	flex-direction: column;
	gap: 11px;
	padding: clamp(22px, 5vw, 36px) clamp(18px, 4.5vw, 32px);
}

.earthcell-content .ec-c6__kind {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 12px;
	letter-spacing: 0.28em;
}

.earthcell-content .ec-c6__kind--satellite {
	color: var(--ec-lime);
}

.earthcell-content .ec-c6__kind--drone {
	color: var(--ec-aqua);
}

.earthcell-content .ec-c6__kind--ground {
	color: #e6eef0;
}

.earthcell-content .ec-c6__statement {
	margin: 0;
	max-width: none;
	font-size: clamp(19px, 3.8vw, 22px);
	line-height: 1.3;
	font-weight: 600;
	color: var(--ec-paper);
}

/*
 * THE TWO HONESTY LINES LIVE IN THIS RULE. NOT YET FLOWN and NO FIELD EVIDENCE RECORDED YET are
 * the sentences that keep the two lower panels from being a claim about work nobody has done.
 * Deleting them, or making them quieter than they are, changes what the chapter says.
 */
.earthcell-content .ec-c6__status {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.12em;
	color: #8c9ba3;
}

/*
 * ONE CREDIT FOR THE THREE PANELS, UNDER THEM AND IN NORMAL FLOW. Inside a panel it would sit over
 * imagery in a corner, and the bottom right corner of the window belongs to the site's chat
 * button. The right padding from .ec-credit is kept and the left one lines the credit up with the
 * reading column above the chapter.
 */
.earthcell-content .ec-credit--c6 {
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: 18px var(--ec-gutter) 30px;
	padding-right: 96px;
}

/* --------------------------------------------------------------------------
   ONE FOCUS STATE FOR EVERYTHING ON THE PAGE.

   THE STYLESHEET HAD ONE :focus RULE IN IT. A visitor moving through this page with a keyboard had
   almost nothing telling them where they were, and the page sits under a B Corp badge. Found in
   the design review of 29 August 2026.

   :focus-visible AND NOT :focus, so a mouse click does not leave a ring behind on a button that
   was only pressed. Browsers decide when it applies and they get it right.

   TWO COLOURS, BECAUSE THE PAGE HAS TWO GROUNDS. Lime reads on the near-black chapters and is
   nearly invisible on paper, so the light sections take teal. The outline is 2px with 2px of
   offset in both cases, so the shape of the marker never changes.
   -------------------------------------------------------------------------- */

.earthcell-content a:focus-visible,
.earthcell-content button:focus-visible,
.earthcell-content summary:focus-visible,
.earthcell-content input:focus-visible,
.earthcell-content textarea:focus-visible {
	outline: 2px solid var(--ec-teal);
	outline-offset: 2px;
}

.earthcell-content .ec-c1 a:focus-visible,
.earthcell-content .ec-c1 button:focus-visible,
.earthcell-content .ec-c3 a:focus-visible,
.earthcell-content .ec-c3 button:focus-visible,
.earthcell-content .ec-c4 a:focus-visible,
.earthcell-content .ec-c4 button:focus-visible,
.earthcell-content .ec-c5 a:focus-visible,
.earthcell-content .ec-c5 button:focus-visible,
.earthcell-content .ec-c6 a:focus-visible,
.earthcell-content .ec-c6 button:focus-visible,
.earthcell-content .ec-c8 a:focus-visible,
.earthcell-content .ec-c8 button:focus-visible,
.earthcell-content .ec-c8 input:focus-visible,
.earthcell-content .ec-c8 textarea:focus-visible {
	outline-color: var(--ec-lime);
}

/*
 * AND A HOVER STATE ON THE THINGS THAT ARE PRESSED. The stylesheet declared transitions on
 * buttons that had nothing to transition to, which is the same thing as having no hover at all.
 */
.earthcell-content .ec-button:hover {
	filter: brightness(1.08);
}

.earthcell-content .ec-c5__year:hover,
.earthcell-content .ec-c8__choice:hover {
	border-color: rgba(206, 241, 123, 0.55);
}

/* ==========================================================================
   THE DISCLOSURES. Native <details>, everywhere on this page, and never rebuilt in JavaScript.

   A NATIVE DISCLOSURE OPENS WITH NO SCRIPT, IS REACHED BY A BROWSER'S OWN FIND ON PAGE IN CURRENT
   BROWSERS, AND IS KEYBOARD OPERABLE WITHOUT ANYBODY WRITING THAT. Rebuilding it in JavaScript
   loses all three and gains nothing. The design handoff says the same thing in its own words.

   THE MARKER HAS TO BE SUPPRESSED TWICE. Chrome and Safari want ::-webkit-details-marker and
   Firefox wants ::marker on the summary. Missing either leaves a browser drawing its own triangle
   beside our plus sign.

   THE PLUS IS TWO BARS ON ONE ELEMENT AND ROTATES 45 DEGREES WHEN THE ANSWER OPENS. It is drawn
   with the two pseudo-elements of a single empty span, so there is no run of inline children for
   the page builder to wrap in a paragraph of its own.

   TWO NAMES, ONE PATTERN. .ec-fold is the small monospace expandable inside a chapter (C2 and C7).
   .ec-faq is the full-width question in C9. They differ only in type size and rule weight.
   ========================================================================== */

.earthcell-content .ec-fold__summary,
.earthcell-content .ec-faq__summary {
	display: flex;
	align-items: center;
	justify-content: space-between;
	gap: 24px;
	list-style: none;
	cursor: pointer;
}

.earthcell-content .ec-fold__summary::-webkit-details-marker,
.earthcell-content .ec-faq__summary::-webkit-details-marker {
	display: none;
}

.earthcell-content .ec-fold__summary::marker,
.earthcell-content .ec-faq__summary::marker {
	content: '';
}

.earthcell-content .ec-fold__plus,
.earthcell-content .ec-faq__plus {
	position: relative;
	flex: none;
	transition: transform 0.18s ease;
}

.earthcell-content .ec-fold__plus::before,
.earthcell-content .ec-fold__plus::after,
.earthcell-content .ec-faq__plus::before,
.earthcell-content .ec-faq__plus::after {
	content: '';
	position: absolute;
	background: currentcolor;
}

.earthcell-content .ec-fold[open] .ec-fold__plus,
.earthcell-content .ec-faq[open] .ec-faq__plus {
	transform: rotate(45deg);
}

/* The small expandable inside a chapter. Monospace, teal, rules above and below. */
.earthcell-content .ec-fold {
	border-top: 1px solid rgba(62, 68, 84, 0.18);
	border-bottom: 1px solid rgba(62, 68, 84, 0.18);
}

.earthcell-content .ec-fold__summary {
	padding: 18px 0;
	font-family: var(--ec-mono);
	font-size: 11px;
	letter-spacing: 0.14em;
	color: var(--ec-teal);
}

.earthcell-content .ec-fold__plus {
	width: 14px;
	height: 14px;
}

.earthcell-content .ec-fold__plus::before {
	top: 6px;
	left: 0;
	width: 14px;
	height: 1.5px;
}

.earthcell-content .ec-fold__plus::after {
	left: 6px;
	top: 0;
	width: 1.5px;
	height: 14px;
}

.earthcell-content .ec-fold__body {
	display: flex;
	flex-direction: column;
	gap: 16px;
	padding: 0 0 22px;
}

.earthcell-content .ec-fold__body > p {
	margin: 0;
	max-width: var(--ec-measure);
	font-size: 17px;
	line-height: 1.65;
}

/* ==========================================================================
   C2 WHY EARTHCELL. A spine of four steps beside one short argument.

   THIS IS THE CHAPTER THAT MOST OFTEN COMES OUT AS TWO BLOCKS OF PROSE, and the design says so.
   It is a spine of four steps beside one short argument, and the steps are the half a reader
   remembers.

   THE LAST STEP IS PERMANENTLY DIMMED AND MUST STAY DIMMED. Certification is optional and it has
   not happened. Drawing it at the weight of the other three promises it, and CLAUDE.md is
   explicit that we never imply a certification that does not exist.
   ========================================================================== */

.earthcell-content .ec-c2 {
	margin-top: 88px;
}

.earthcell-content .ec-c2__inner {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 300px), 1fr));
	gap: clamp(30px, 5vw, 72px);
	align-items: start;
}

.earthcell-content .ec-c2__words {
	display: flex;
	flex-direction: column;
	gap: 28px;
	max-width: var(--ec-measure);
}

.earthcell-content .ec-c2__title {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(28px, 11.6px + 3.16vw, 52px);
	line-height: 1.04;
	font-weight: 800;
	letter-spacing: -0.03em;
}

.earthcell-content .ec-c2__body {
	margin: 0;
	max-width: none;
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
}

/*
 * THE SPINE IS A LEFT RULE ON THE LIST, NOT A BORDER ON EACH STEP. One line down the whole
 * sequence is what makes it read as an order of work rather than as four separate notes.
 */
.earthcell-content .ec-c2__spine {
	margin: 0;
	padding: 0 0 0 0;
	max-width: none;
	list-style: none;
	border-left: 1.5px solid rgba(62, 68, 84, 0.16);
}

.earthcell-content .ec-c2__step {
	margin: 0;
	padding: 22px 0 22px 28px;
	list-style: none;
}

.earthcell-content .ec-c2__step::before,
.earthcell-content .ec-c2__step::marker {
	content: none;
}

.earthcell-content .ec-c2__step-label {
	margin: 0 0 8px;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.18em;
	color: var(--ec-teal);
}

.earthcell-content .ec-c2__step-title {
	margin: 0;
	max-width: none;
	font-size: clamp(20px, 4.6vw, 26px);
	line-height: 1.18;
	font-weight: 700;
	letter-spacing: -0.02em;
}

/* Certification, if ever. Dimmed on purpose, label and title together. */
.earthcell-content .ec-c2__step--last .ec-c2__step-label,
.earthcell-content .ec-c2__step--last .ec-c2__step-title {
	color: rgba(62, 68, 84, 0.55);
}

/* ==========================================================================
   C7 EVIDENCE LADDER. Three rungs, and the third one is dashed.

   THE DASHED RUNG IS THE ARGUMENT, NOT A STYLE. Observation happens, a project record is kept,
   and formal verification is optional, has not happened and may never apply. A solid third rung
   changes what the chapter says.
   ========================================================================== */

.earthcell-content .ec-c7 {
	margin-top: 88px;
}

.earthcell-content .ec-c7__inner {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 300px), 1fr));
	gap: clamp(30px, 5vw, 72px);
	align-items: start;
}

.earthcell-content .ec-c7__words {
	display: flex;
	flex-direction: column;
	gap: 24px;
	max-width: var(--ec-measure);
}

.earthcell-content .ec-c7__title {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(26px, 12.3px + 2.63vw, 46px);
	line-height: 1.06;
	font-weight: 800;
	letter-spacing: -0.03em;
}

.earthcell-content .ec-c7__body {
	margin: 0;
	max-width: none;
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
}

.earthcell-content .ec-c7__ladder {
	display: flex;
	flex-direction: column;
	margin: 0;
	padding: 0;
	max-width: none;
	list-style: none;
}

.earthcell-content .ec-c7__rung {
	margin: 0;
	padding: 20px 22px;
	list-style: none;
	border: 1.5px solid var(--ec-teal);
}

.earthcell-content .ec-c7__rung::before,
.earthcell-content .ec-c7__rung::marker {
	content: none;
}

/*
 * THE CONNECTORS ARE A PSEUDO-ELEMENT ON EACH RUNG AFTER THE FIRST, so the rungs read as one
 * sequence. 28px tall, centred, and they take their own space rather than overlapping a border.
 */
.earthcell-content .ec-c7__rung + .ec-c7__rung {
	margin-top: 28px;
	position: relative;
}

.earthcell-content .ec-c7__rung + .ec-c7__rung::after {
	content: '';
	position: absolute;
	left: 50%;
	top: -29px;
	width: 1.5px;
	height: 28px;
	background: rgba(0, 90, 105, 0.5);
}

/* Step two is the one that already happens on this project, so it is the one that is filled. */
.earthcell-content .ec-c7__rung--record {
	background: rgba(206, 241, 123, 0.22);
}

/*
 * ONLY IF IT APPLIES. Dashed and dimmed, and both are the meaning rather than the decoration.
 */
.earthcell-content .ec-c7__rung--maybe {
	border: 1.5px dashed rgba(62, 68, 84, 0.35);
}

.earthcell-content .ec-c7__rung--maybe .ec-c7__rung-label,
.earthcell-content .ec-c7__rung--maybe .ec-c7__rung-title,
.earthcell-content .ec-c7__rung--maybe .ec-c7__rung-note {
	color: rgba(62, 68, 84, 0.62);
}

.earthcell-content .ec-c7__rung-label {
	margin: 0 0 8px;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.18em;
	color: var(--ec-teal);
}

.earthcell-content .ec-c7__rung-title {
	margin: 0 0 6px;
	max-width: none;
	font-size: clamp(19px, 3.8vw, 22px);
	line-height: 1.2;
	font-weight: 700;
}

.earthcell-content .ec-c7__rung-note {
	margin: 0;
	max-width: none;
	font-size: 15px;
	line-height: 1.5;
}

/* ==========================================================================
   C9 FOUR QUESTIONS. Full-width native disclosures on the light ground.

   EACH ONE ANSWERS SOMETHING A CORPORATE READER ASSUMES, and three of the four are assumptions
   CLAUDE.md forbids letting stand: that land is being bought, that this is a carbon credit, and
   that the money is spent inside the lines on the map.
   ========================================================================== */

.earthcell-content .ec-c9 {
	margin-top: 88px;
}

.earthcell-content .ec-c9__inner {
	max-width: var(--ec-measure);
}

.earthcell-content .ec-faq {
	border-top: 1px solid rgba(62, 68, 84, 0.18);
}

/* The last one closes the stack, so the four read as one block rather than as four loose rules. */
.earthcell-content .ec-faq--last {
	border-bottom: 1px solid rgba(62, 68, 84, 0.18);
}

.earthcell-content .ec-faq__summary {
	padding: 22px 0;
	font-size: clamp(17px, 4.2vw, 20px);
	line-height: 1.3;
	font-weight: 700;
	color: var(--ec-ink);
}

.earthcell-content .ec-faq__plus {
	width: 18px;
	height: 18px;
	color: var(--ec-teal);
}

.earthcell-content .ec-faq__plus::before {
	top: 8px;
	left: 0;
	width: 18px;
	height: 1.5px;
}

.earthcell-content .ec-faq__plus::after {
	left: 8px;
	top: 0;
	width: 1.5px;
	height: 18px;
}

.earthcell-content .ec-faq__answer > p {
	margin: 0;
	max-width: var(--ec-measure);
	padding-bottom: 24px;
	font-size: clamp(16px, 2.4vw, 18px);
	line-height: 1.65;
}

/* ==========================================================================
   C8 SUPPORT. The page's ask, and the only part of it that talks back to the map.

   THE THREE OPTIONS ARE A LIST, NOT A SET OF BOXES. Each one carries only a top rule, full width,
   and the chosen one takes a background. Boxing them would make three cards competing with the
   enquiry panel beside them, and the panel is the thing the page wants pressed.

   THE SIXTEEN SQUARES ARE IDENTICAL IN THE MARKUP AND THE STYLESHEET COLOURS THE RIGHT ONES.
   The page ships sixteen empty divs per option, and the rules below light square 6 for the first
   two and squares 2, 3, 6, 7, 10 and 11 for the third. A change to which squares light is a change
   to one selector here rather than to forty-eight elements in a WordPress page body.

   THE CHOSEN OPTION IS AN ATTRIBUTE ON THE SECTION, written by the widget. With no script there is
   no attribute, and the state with no attribute is three equal options and none of them lit, which
   is honest rather than broken.

   NO BREAKPOINTS. The two columns are minmax(min(100%, 300px), 1fr) and collapse when they stop
   fitting. The chips wrap on flex.
   ========================================================================== */

.earthcell-content .ec-c8 {
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: 104px;
	background: var(--ec-ground);
	color: #e6eef0;
}

.earthcell-content .ec-c8__inner {
	display: flex;
	flex-direction: column;
	gap: 26px;
	max-width: var(--ec-band);
	margin: 0 auto;
	padding: clamp(56px, 9vw, 96px) var(--ec-gutter);
}

/* The largest heading on the page, because it is the page's ask. */
.earthcell-content .ec-c8__title {
	margin: 0;
	max-width: var(--ec-heading);
	font-size: clamp(28px, 11.6px + 3.16vw, 52px);
	line-height: 1.04;
	font-weight: 800;
	letter-spacing: -0.03em;
	color: var(--ec-paper);
}

.earthcell-content .ec-c8__body {
	margin: 0;
	max-width: var(--ec-measure);
	font-size: clamp(16px, 2.5vw, 19px);
	line-height: 1.65;
	color: rgba(230, 238, 240, 0.82);
}

/*
 * ONE COLUMN, NOT TWO. Rebuilt 30 August 2026.
 *
 * It was a two-up grid: the scale options on one side and the enquiry panel on the other, which
 * made them read as two alternatives rather than as one ask with an optional field in it. The
 * chapter is now the panel, and beneath the panel a quiet second route to a cell record.
 *
 * The width is the prose measure rather than the band, because a form wider than the sentences
 * above it looks like a different page.
 */
.earthcell-content .ec-c8__grid {
	display: flex;
	flex-direction: column;
	gap: clamp(30px, 4vw, 40px);
	max-width: var(--ec-measure);
	margin-top: 18px;
}

/*
 * THREE COMPACT CHIPS IN A ROW, NOT THREE STACKED LIST ROWS.
 *
 * They used to be full-width rows with a title and a sentence each, which was right when the scale
 * was the chapter's subject. It is an optional field inside the enquiry now, so the three sit
 * side by side and carry a glyph and a title only.
 *
 * ONE COLUMN BELOW 1160px, THREE ABOVE IT, and never two. Same rule and same reason as the partner
 * cards further up this file: a per-component auto-fit minimum produces orphan rows.
 */
.earthcell-content .ec-c8__scales {
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	gap: 8px;
}

@media (min-width: 1160px) {
	.earthcell-content .ec-c8__scales {
		grid-template-columns: repeat(3, minmax(0, 1fr));
	}
}

/*
 * The optional-scale label is quieter than the panel's own heading, which is the lime one.
 *
 * THE TOP MARGIN IS THE PANEL'S OWN RHYTHM AND NOT A NUDGE. The panel spaces its children with
 * their own margins rather than with a gap, and the sentence above this one sets no bottom margin,
 * so at 4px the label sat directly on the last line of that paragraph and read as part of it.
 *
 * BOTH CLASSES ARE NAMED ON PURPOSE. The base label rule sets the "margin" shorthand and it appears
 * later in this file, so at equal specificity it won and this margin did nothing at all. Measured
 * on staging: the gap above the label was still zero after the rule had been deployed.
 */
.earthcell-content .ec-c8__label.ec-c8__label--optional {
	margin-top: clamp(22px, 3vw, 28px);
	color: #8c9ba3;
}

/*
 * THE QUIET ROUTE TO A RECORD, under the ask and separated from it by a rule.
 *
 * It is not a second call to action and it must not look like one. A hairline above it, the same
 * muted label as everything else at this size, and no button of any colour.
 */
.earthcell-content .ec-c8__records {
	padding-top: clamp(18px, 3vw, 24px);
	border-top: 1px solid rgba(230, 238, 240, 0.12);
}

.earthcell-content .ec-c8__label {
	margin: 0 0 14px;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.18em;
	color: #8c9ba3;
}

/*
 * A COMPACT CHIP, NOT A LIST ROW. Changed 30 August 2026.
 *
 * IT USED TO BE A FULL-WIDTH ROW with the cell geometry in a 40px first column, a title and a
 * sentence, separated from its neighbours by a hairline. That was the right shape while choosing a
 * scale was the chapter's subject. It is an optional field inside the enquiry now, so three of
 * these sit side by side and each carries a glyph and a title.
 *
 * A REAL BOX WITH A REAL BORDER, because it is a control a visitor picks one of, and a row with
 * only a top rule reads as a list of things to read rather than a set of things to choose.
 */
.earthcell-content .ec-c8__choice {
	display: flex;
	align-items: center;
	gap: 10px;
	width: 100%;
	padding: 12px 14px;
	background: transparent;
	border: 1px solid rgba(230, 238, 240, 0.16);
	border-radius: 0;
	text-align: left;
	cursor: pointer;
	transition: background 0.25s ease, border-color 0.25s ease;
}

.earthcell-content .ec-c8[data-ec-scale-chosen="0"] .ec-c8__choice--part,
.earthcell-content .ec-c8[data-ec-scale-chosen="1"] .ec-c8__choice--one,
.earthcell-content .ec-c8[data-ec-scale-chosen="2"] .ec-c8__choice--larger {
	background: rgba(206, 241, 123, 0.1);
}

/* --------------------------------------------------------------------------
   THE SCALE GLYPHS ARE HEXAGONS, AND THEY WERE SQUARES UNTIL 29 AUGUST 2026.

   Round 4 of the design handoff, in its own words: "The honeycomb is the page's shape now, so the
   two diagram sets follow it. Do not put squares back anywhere: the product is called EarthCell
   and the cells are cells." The map above draws 799 hexagons. A row of little squares underneath
   claiming to show the same thing at three scales was the page disagreeing with itself.

   EIGHT BY NINE PIXELS, WHICH IS A REGULAR HEXAGON AT THIS SIZE. A pointy-top hexagon of
   circumradius R is 2R tall and R times the square root of three wide. At R of 4.5 that is 9 by
   7.79, and 8 is the nearest whole pixel. Getting this backwards gives a hexagon lying on its
   side, which tiles correctly and is rotated ninety degrees from every other cell on the page.

   THEY ARE FILLED RATHER THAN OUTLINED BECAUSE clip-path REMOVES A BORDER. A border is drawn on
   the element's own rectangle and the clip then cuts the corners off it, so an outlined hexagon
   comes out as four disconnected fragments of line. The four fills below are round 4's own.

   EVERY OTHER ROW SHIFTS HALF A COLUMN AND EVERY ROW OVERLAPS THE ONE ABOVE. Those two together
   are what make a honeycomb rather than a grid of hexagons with diamond holes between them. The
   negative bottom margin is a quarter of the height, which is the overlap a pointy-top lattice
   needs, and the same ratio the map's own rows use.
   -------------------------------------------------------------------------- */

/*
 * "grid-row: span 2" IS GONE, and it has to be. The chip was a two-row grid with the glyph spanning
 * both rows beside a title and a sentence. It is a flex row with a glyph and a title now, and a row
 * span inside a flex parent is ignored, so leaving it in would have been a line that looked
 * meaningful and did nothing.
 */
.earthcell-content .ec-c8__cells {
	display: grid;
	grid-template-columns: repeat(4, 8px);
	gap: 1px;
	flex: none;
	/* Four columns and the half-column shift the odd rows take. */
	width: 39px;
	/*
	 * THE ROWS MUST NOT STRETCH, AND THEY DID.
	 *
	 * This element spans both rows of the option's own grid, so it is as tall as the title and the
	 * body together. A grid's rows default to stretching into whatever height they are given, which
	 * pulled the four rows apart to 13px each and undid the overlap that makes a honeycomb. Packing
	 * them to the start keeps each row at its own 7px and lets the container be taller than they
	 * are, which is what it is for.
	 */
	align-content: start;
}

.earthcell-content .ec-c8__cell {
	width: 8px;
	height: 9px;
	/* A quarter of the height, so each row sits into the one above it. */
	margin-bottom: -2px;
	border: 0;
	clip-path: polygon(50% 0, 100% 25%, 100% 75%, 50% 100%, 0 75%, 0 25%);
	background: rgba(206, 241, 123, 0.13);
}

/*
 * ROWS TWO AND FOUR, WHICH ARE CHILDREN 5 TO 8 AND 13 TO 16. Counting them out rather than using
 * a formula because sixteen cells in four rows is small enough to read, and a reader checking
 * which cells light further down is already counting children.
 */
.earthcell-content .ec-c8__cell:nth-child(n + 5):nth-child(-n + 8),
.earthcell-content .ec-c8__cell:nth-child(n + 13) {
	transform: translateX(50%);
}

/*
 * WHICH CELLS LIGHT, AND WHY THEY ARE THE ONES THEY ARE.
 *
 * Cell 6 is the second of the second row, which is the middle of a four by four block and reads as
 * one cell inside a landscape rather than a corner of it. The larger option lights a block of six,
 * so the three rows say the same thing at three scales: part of a cell, one cell, a group.
 *
 * THE THREE FILLS ARE ROUND 4'S OWN. Aqua marks the part-cell, because aqua is this page's colour
 * for the thing that is not a whole one. Lime is a funded cell. The dimmer lime is a region: a
 * group of cells somebody is supporting together rather than six separately funded ones.
 *
 * THE FIRST ONE IS NO LONGER HALF FILLED. A half fill was a gradient across the element, and
 * clip-path cuts the hexagon out of that rectangle afterwards, so the join landed across the
 * hexagon's waist at an angle that read as a rendering fault rather than as half a cell. Aqua on
 * its own says the same thing, and it is the colour the rest of the page already uses for it.
 */
.earthcell-content .ec-c8__choice--part .ec-c8__cell:nth-child(6) {
	background: var(--ec-aqua);
}

.earthcell-content .ec-c8__choice--one .ec-c8__cell:nth-child(6) {
	background: var(--ec-lime);
}

.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(2),
.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(3),
.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(6),
.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(7),
.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(10),
.earthcell-content .ec-c8__choice--larger .ec-c8__cell:nth-child(11) {
	background: rgba(206, 241, 123, 0.4);
}

/*
 * SMALLER, because the chip is a field label now rather than a heading.
 *
 * At 22px, three of these side by side read as three offers competing with the button below them.
 * The scale is optional and the ask is not, so the ask is the biggest thing in the panel.
 */
.earthcell-content .ec-c8__choice-title {
	margin: 0;
	max-width: none;
	font-size: clamp(15px, 2vw, 16px);
	line-height: 1.3;
	font-weight: 700;
	color: var(--ec-paper);
}

.earthcell-content .ec-c8__choice-body {
	margin: 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.5;
	color: rgba(230, 238, 240, 0.72);
}

/*
 * THE ONE LINE SAYING WHAT THIS COSTS. It sits directly under the three scale options because that
 * is where a reader looks for a number and does not find one. Body size rather than caption size:
 * it is a sentence a funder needs to read, not a footnote.
 */
.earthcell-content .ec-c8__note {
	margin: 20px 0 34px;
	max-width: var(--ec-measure);
	font-size: 16px;
	line-height: 1.55;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content .ec-c8__chips {
	display: flex;
	flex-wrap: wrap;
	gap: 10px;
}

/*
 * A CHIP FLIES THE MAP. Clicking one goes back to the instrument at the top of the page, moves the
 * camera to that cell and opens its record. The design is explicit that a chip which names a cell
 * and does nothing is a dead end, and it is right.
 */
.earthcell-content .ec-c8__chip {
	display: flex;
	flex-direction: column;
	gap: 5px;
	padding: 12px 14px;
	background: transparent;
	border: 1px solid rgba(230, 238, 240, 0.2);
	border-radius: 0;
	text-align: left;
	cursor: pointer;
	transition: border-color 0.2s ease;
}

.earthcell-content .ec-c8__chip:hover,
.earthcell-content .ec-c8__chip:focus-visible {
	border-color: var(--ec-lime);
}

.earthcell-content .ec-c8__chip-code {
	font-family: var(--ec-mono);
	font-size: 11px;
	letter-spacing: 0.06em;
	color: var(--ec-paper);
}

/*
 * SAMPLE STAYS ON EVERY CHIP. The funding state of a cell comes from the demonstration boundary
 * and no one has funded anything on this project. CLAUDE.md requires that to be marked on screen
 * while it is true.
 */
.earthcell-content .ec-c8__chip-status {
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.14em;
	color: #8c9ba3;
}

/*
 * ONE PRIMARY ACTION IN THIS CHAPTER, AND THERE WERE FOUR.
 *
 * Three scale buttons, three cell chips, a form and a lime "Explore the EarthCells" button, all
 * competing at the same weight. The design review counts that as four peers with no ranking, and
 * round 4 settles it: C8 is one primary action, discussing a corporate contribution, with the
 * scale options feeding it.
 *
 * SO LIME NOW MEANS SEND ENQUIRY AND NOTHING ELSE IN THIS CHAPTER. Going back to the map is a real
 * thing a reader may want, so the link stays; it stops being a button. The rule the page follows
 * everywhere else is that lime is the primary action, and having two of them meant it was not.
 */
.earthcell-content .ec-c8__action .ec-button--lime {
	padding: 0;
	background: transparent;
	color: var(--ec-lime);
	text-decoration: underline;
	text-underline-offset: 5px;
}

.earthcell-content .ec-c8__action .ec-button--lime:hover {
	filter: none;
	color: var(--ec-paper);
}

.earthcell-content .ec-c8__action {
	margin: 30px 0 0;
	max-width: none;
}

/* ==========================================================================
   C10 ABOUT THIS PROTOTYPE. Every caveat on the page lands here and nowhere else.

   THE DESIGN SAYS IT IN THOSE WORDS: "Every caveat on the page lands here. Don't distribute it
   back into the chapters." A warning that appears in four places is a warning a reader learns to
   skip. The demonstration note that used to sit beside "See the land your funding could support"
   is in this section now, and the section it left carries a comment saying where it went.

   A 1.5px RULE ABOVE IT, NOT A BAND. This is the last thing on the page and it is deliberately
   quiet: light ground, small type, a rule to separate it from the ask above it. Making it loud
   would be making the caveats the closing argument of the page, which they are not.
   ========================================================================== */

.earthcell-content .ec-c10 {
	margin-top: 88px;
	padding-bottom: clamp(24px, 4vw, 48px);
}

.earthcell-content .ec-c10__inner {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 300px), 1fr));
	gap: clamp(28px, 4vw, 56px);
	align-items: start;
	padding-top: 36px;
	border-top: 1.5px solid rgba(62, 68, 84, 0.16);
}

.earthcell-content .ec-c10__words {
	display: flex;
	flex-direction: column;
	gap: 14px;
	max-width: var(--ec-measure);
}

.earthcell-content .ec-c10__words > p {
	margin: 0;
	max-width: none;
	font-size: clamp(16px, 2.2vw, 17px);
	line-height: 1.65;
}

.earthcell-content .ec-c10__eyebrow {
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.18em;
	color: var(--ec-teal);
}

.earthcell-content .ec-c10__stack {
	display: flex;
	flex-direction: column;
	border: 1px solid rgba(62, 68, 84, 0.2);
}

.earthcell-content .ec-c10__row {
	display: flex;
	flex-direction: column;
	gap: 5px;
	padding: 16px 18px;
	border-bottom: 1px solid rgba(62, 68, 84, 0.14);
}

/*
 * The last row is the one the page is about, so it is the one that is lit, and it also drops the
 * dividing rule because there is nothing under it to divide from.
 */
.earthcell-content .ec-c10__row--lit {
	border-bottom: 0;
	background: rgba(206, 241, 123, 0.22);
}

.earthcell-content .ec-c10__label {
	margin: 0;
	font-family: var(--ec-mono);
	font-size: 10px;
	letter-spacing: 0.16em;
	color: var(--ec-teal);
}

.earthcell-content .ec-c10__line {
	margin: 0;
	font-size: 15px;
	line-height: 1.4;
}

/*
 * KEPT, THOUGH NOTHING USES IT NOW. The empty 4/3 frame with the note standing in it, for the
 * next slot that has a picture missing. The teal band's own slot has its photograph.
 */
.earthcell-content .ec-note--image-slot {
	display: flex;
	align-items: flex-end;
	aspect-ratio: 4 / 3;
}

/* --------------------------------------------------------------------------
   The photograph, and the line that says what it is.

   THE CAPTION IS NOT MARKED AS UNRESOLVED. Everything on this page in a 12px monospace box with
   an aqua border is something nobody has confirmed yet. The picture is confirmed: it is C Level's
   own photograph of this project. So the caption borrows the monospace and the colour, because it
   is a technical string like every other label here, and it does not borrow the border, because
   the border means "this is not settled".

   4/3 AND object-fit: cover, so a picture of another shape cannot stretch the column or leave a
   gap under it. The one in the slot today is 4096 by 3072, which is 4/3 exactly, so nothing is
   cropped off it.
   -------------------------------------------------------------------------- */

/*
 * WORDPRESS WRAPS EVERY HTML COMMENT IN THE PAGE BODY IN A PARAGRAPH OF ITS OWN, and inside a
 * grid that paragraph takes a cell.
 *
 * wpautop sees the blank lines around a comment and makes a block out of it. The comment renders
 * nothing, so seven of these sat on the page unnoticed. Then one landed between the text column
 * and the photograph in the teal band: it took the right-hand cell, and the picture dropped to a
 * second row underneath the words instead of sitting beside them. Reported by Piers within a
 * minute of the deploy, which is how visible it was.
 *
 * :empty matches a paragraph holding only a comment, because a comment is not a child for this
 * purpose. Hiding it takes it out of the grid entirely rather than leaving an invisible cell.
 *
 * The alternative is to stop writing comments in the page body. The comments are how that file
 * explains which sentences are Piers's and which are not, so the stylesheet gives way instead.
 */
.earthcell-content p:empty {
	display: none;
}

/*
 * THERE IS NO RULE HERE FOR <p><br></p>, AND ONE WAS TRIED AND TAKEN OUT THE SAME HOUR.
 *
 * :empty does not match a paragraph holding a line break, because the break is a child, so
 * `p:has(> br:only-child)` looked like the missing guard. It is not. :only-child counts element
 * children and ignores text, so that selector also matches any real two-line paragraph, and it
 * hid the locator's own coordinates caption in C4 within a minute of being deployed.
 *
 * CSS cannot say "a paragraph with no text in it". The artefacts that shape came from are fixed
 * where they start instead: the ring and the dot in C4 are block elements so WordPress never
 * wraps them, and the script tag below has a rule of its own.
 */

/*
 * AND THE PARAGRAPH WORDPRESS WRAPS AROUND THE WIDGET'S OWN SCRIPT TAG.
 *
 * The script is the last thing in the page body and it carries a comment above it, so wpautop
 * makes a paragraph of the pair and it lands as 28px of empty space under the last chapter. The
 * script still runs: CSS decides what is painted and never decides what executes.
 *
 * A separate rule again, for the same reason as the one above it.
 */
.earthcell-content > p:has(> script) {
	display: none;
}

.earthcell-content .ec-figure__image {
	display: block;
	width: 100%;
	height: auto;
	aspect-ratio: 4 / 3;
	object-fit: cover;
}

.earthcell-content .ec-figure__caption {
	margin: 12px 0 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1.6;
	letter-spacing: 0.04em;
	color: var(--ec-aqua);
}

/*
 * The old class name, kept because it is the one the theme has always exposed to page content and
 * something else may still be using it.
 */
.earthcell-note {
	display: block;
	max-width: var(--ec-measure);
	margin: 1.5rem 0;
	padding: 0.9rem 1.1rem;
	border-left: 3px solid var(--ec-aqua);
	background: #f0fbfa;
	font-family: var(--ec-mono);
	font-size: 0.85rem;
	line-height: 1.6;
	color: #12525c;
}

/* --------------------------------------------------------------------------
   Monospace markers. Every technical string on the page uses this stack, the same one the map's
   own labels use (section 11). Nothing here is a new webfont.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-marker {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.16em;
	color: var(--ec-teal);
}

/* --------------------------------------------------------------------------
   "A supporter could return and see", as seven things rather than one sentence.

   THE WORDS ARE UNCHANGED. The page ran them as a single sentence with six commas in it, which is
   a list nobody reads. Section 0B.6 asks for the same words, scannable.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-returns {
	max-width: 380px;
}

.earthcell-content .ec-returns__list {
	margin: 14px 0 0;
	padding: 0;
	max-width: none;
	list-style: none;
}

.earthcell-content .ec-returns__list li {
	margin: 0;
	padding: 9px 0;
	border-bottom: 1px solid rgba(62, 68, 84, 0.14);
	font-size: 16px;
	line-height: 1.5;
}

.earthcell-content .ec-node__label {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.18em;
	color: var(--ec-lime);
}

.earthcell-content .ec-node__value {
	margin: 6px 0 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.4;
}

/* The cell that the whole model hangs on, so it is the one node drawn in the accent. */
.earthcell-content .ec-node--object {
	border: 1.5px solid var(--ec-lime);
	background: rgba(206, 241, 123, 0.08);
}

.earthcell-content .ec-node--object .ec-node__value {
	font-family: var(--ec-mono);
	font-size: 15px;
	letter-spacing: 0.06em;
}

/*
 * EC-KOGI-000027 is a code from a demonstration grid over a demonstration boundary. Nobody has
 * been issued an EarthCell on this project, so the code carries the marker the design uses for
 * every other sample value on the page.
 */
.earthcell-content .ec-node__marker {
	margin: 8px 0 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.16em;
	color: var(--ec-aqua);
}

.earthcell-content .ec-node--streams {
	display: grid;
	grid-template-columns: repeat(4, minmax(0, 1fr));
	gap: 18px;
	max-width: none;
	text-align: left;
}

.earthcell-content .ec-stream__kind {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 9px;
	line-height: 1.6;
	letter-spacing: 0.16em;
	color: var(--ec-aqua);
}

.earthcell-content .ec-stream__label {
	margin: 6px 0 0;
	max-width: none;
	font-size: 13px;
	line-height: 1.4;
	color: rgba(230, 238, 240, 0.82);
}

/* --------------------------------------------------------------------------
   The four monitoring readouts beside the diagram.

   THEY SAY WHAT HAS NOT HAPPENED. Two of the four are a zero and the other two say no observation
   has been recorded, which is the honest state of this project and the reason the readouts are
   worth showing at all. The SAMPLE notes on the last two are demonstration markers and must not
   be removed.

   THE NOTE COLOUR IS NOT IN THE DESIGN FILE. Section 0B.6 asks for "a 9px grey note" and section
   11 lists no grey token; the two greys the design file does use appear only on its dark bands
   and are too light to read on paper. This is the ink token held back, which keeps the palette
   closed.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-readouts {
	max-width: 300px;
}

.earthcell-content .ec-readout__label {
	margin: 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 9px;
	line-height: 1.6;
	letter-spacing: 0.16em;
	color: var(--ec-teal);
}

.earthcell-content .ec-readout__value {
	margin: 8px 0 0;
	max-width: none;
	font-size: 22px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: -0.02em;
}

.earthcell-content .ec-readout__note {
	margin: 6px 0 0;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 9px;
	line-height: 1.6;
	letter-spacing: 0.12em;
	/*
	 * SOLID INK, NOT INK AT 62 PER CENT.
	 *
	 * At rgba(62,68,84,.62) on the paper these measured 3.31:1, under the 4.5:1 the standard asks
	 * of text this size. They are the notes carrying the SAMPLE markers, so they are the last text
	 * on the page that should be hard to read. The full token measures about 8.9:1.
	 *
	 * This page already made the same fix once, when a 10px notice over bright imagery was raised
	 * to 11px. Faint grey on pale paper is the same fault wearing different clothes.
	 */
	color: var(--ec-ink, #3e4454);
}

/* --------------------------------------------------------------------------
   The enquiry panel.

   IT IS PRESENTATIONAL AND IT SAYS SO ON SCREEN. No endpoint is wired to these three fields and
   the design never asked for one. The enquiry route that does work is the one inside the map, on
   the component's own data-enquiry-endpoint. A form that looks live and swallows what somebody
   types is worse than no form, so the marker at the foot of the panel is not decoration and must
   not be removed while the button does nothing.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-panel {
	padding: 32px;
	border: 1px solid rgba(230, 238, 240, 0.18);
}

/*
 * LIME, NOT GREY. This is the panel's own heading and the only thing in the chapter that says what
 * the panel is for. Grey put it at the same weight as the optional-scale label below it, so the
 * ask and the optional field read as two equal steps. The spec sets it lime for that reason.
 */
.earthcell-content .ec-marker--panel {
	letter-spacing: 0.18em;
	color: var(--ec-lime);
}

/*
 * 16px AND NOT 14px AT 62 PER CENT. This sentence carries the two facts a corporate reader needs
 * before they type anything: that contributions are agreed individually, and that sending this is
 * not a commitment. It was set as small print, which is the wrong weight for the only answer on
 * the page to "what will this cost me".
 */
.earthcell-content .ec-panel__note {
	margin: 14px 0 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.55;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content .ec-field {
	margin-top: 16px;
}

.earthcell-content .ec-field__label {
	display: block;
	margin-bottom: 6px;
	font-size: 14px;
	line-height: 1.4;
	color: rgba(230, 238, 240, 0.82);
}

.earthcell-content .ec-field__input {
	display: block;
	width: 100%;
	min-height: 48px;
	padding: 12px 14px;
	background: rgba(230, 238, 240, 0.06);
	border: 1px solid rgba(230, 238, 240, 0.22);
	color: #e6eef0;
	font-family: inherit;
	font-size: 16px;
	line-height: 1.4;
}

.earthcell-content .ec-field__input--area {
	resize: vertical;
}

.earthcell-content .ec-panel__action {
	margin: 20px 0 0;
	max-width: none;
}

.earthcell-content .ec-marker--demo {
	margin: 16px 0 0;
	letter-spacing: 0.06em;
	color: #8c9ba3;
}

/* --------------------------------------------------------------------------
   Narrower windows.

   THE FIXED COLUMNS GO FIRST. A 380px or 300px column beside prose stops working long before a
   phone: at about 900px the prose column beside it is narrower than the caveat standing next to
   it, which reads as two sidebars and no article. Below that everything is one column.
   -------------------------------------------------------------------------- */

@media (max-width: 900px) {
	.earthcell-content .ec-grid--sidebar,
	.earthcell-content .ec-grid--readouts,
	.earthcell-content .ec-grid--pair,
	.earthcell-content .ec-contrast {
		grid-template-columns: minmax(0, 1fr);
		gap: 40px;
	}

	.earthcell-content .ec-returns,
	.earthcell-content .ec-readouts {
		max-width: var(--ec-measure);
	}

	.earthcell-content .ec-section,
	.earthcell-content .ec-cta,
	.earthcell-content .ec-band {
		margin-top: 64px;
	}

	.earthcell-content .ec-band__inner {
		padding: 64px var(--ec-gutter);
	}
}

@media (max-width: 780px) {

	.earthcell-content .ec-band__inner {
		padding: 56px var(--ec-gutter);
	}

	.earthcell-content .ec-node--streams {
		grid-template-columns: repeat(2, minmax(0, 1fr));
	}

	.earthcell-content .ec-panel {
		padding: 24px 20px;
	}

	.earthcell-content .ec-intro .ec-lead {
		font-size: 19px;
	}

	.earthcell-content .ec-intro > p {
		font-size: 17px;
	}

	.earthcell-content .ec-col > p {
		font-size: 17px;
	}
}

/* ==========================================================================
   PIERS'S COPY OF 28 AUGUST 2026, SECOND SET. The blocks it needed that did not exist.

   His second copy set replaced the whole page below the map and brought four shapes with it: three
   principles, three named Kogi sites, a strapline that carries a section on its own, and a closing
   line under a two-column block. These are those four and nothing else. Everything above this
   section still does the work it always did.

   NO BREAKPOINTS, WHICH IS THE ROUND 3 RULE FOR EVERYTHING BELOW THE MAP. Every grid here is
   repeat(auto-fit, minmax(min(100%, Npx), 1fr)). The min(100%, ...) is the part that stops the
   page overflowing sideways at 320px, so it stays. Columns collapse when their content stops
   fitting and no width is named anywhere.
   ========================================================================== */

/*
 * THE STRAPLINE. "Fund a real place. Understand who cares for it. Follow what happens next." and
 * "Precise geography. Repeat observation. Time-stamped evidence. Clear stewardship."
 *
 * Both are four short sentences doing the job of a heading. At body size they read as another
 * paragraph and the reader goes past them. They are the two lines on the page most worth stopping
 * on, so they are set larger and tighter, in the ink colour rather than the muted one.
 */
/*
 * THE BOTTOM MARGIN IS NOT ZERO ANY MORE, AND THAT WAS AN UNHANDLED CASE RATHER THAN A CHOICE.
 *
 * This line is used twice. Under the introduction it is followed by a call to action that brings
 * its own 20px of space, so nothing showed. Inside "Digital monitoring, observation and
 * visibility" it is followed by a paragraph whose top margin is zero, and the two touched with no
 * separation at all. Found in the design review of 29 August 2026.
 */
.earthcell-content .ec-intro__strapline {
	margin: 28px 0 20px;
	max-width: 30ch;
	font-size: clamp(20px, 3.4vw, 26px);
	line-height: 1.3;
	font-weight: 700;
	letter-spacing: -0.02em;
	color: var(--ec-teal);
}

/*
 * A closing line UNDER a block that had columns above it.
 *
 * Three of his sections end with a sentence that belongs to the whole section rather than to the
 * left column of it. Without this they inherited the column's width and sat hard against whatever
 * was above, which read as a stray paragraph rather than as the section's own last word.
 */
.earthcell-content .ec-col--after {
	margin-top: clamp(28px, 4vw, 44px);
}

/* --------------------------------------------------------------------------
   Three simple principles.

   THREE EQUAL COLUMNS THAT ARE NOT BOXES. The design's own three-up blocks carry a rule across the
   top rather than a border all the way round, because a boxed grid of three reads as a pricing
   table and this is a list of principles. 240px is the narrowest a title plus two lines of body
   stays readable at, so below about 800px they become three stacked blocks on their own.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-trio {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 240px), 1fr));
	gap: clamp(24px, 3.5vw, 44px);
	margin-top: clamp(32px, 4.5vw, 52px);
}

.earthcell-content .ec-trio__item {
	padding-top: 20px;
	border-top: 1.5px solid var(--ec-teal);
}

/* The rule on the dark ground, where teal is nearly invisible.
   The only place ec-trio is used today is the C3 section, which sits on --ec-ground (#05080b).
   Teal against that ground is about 2.6:1, so the line separating the three principles reads as
   a smudge, and the three run together as one block of text. The design draws it in --ec-lime,
   which is what the design's own three-up blocks use on a dark section. About 15:1.
   The teal rule above stays as the default, because a lime rule on a pale section would
   disappear the same way in the opposite direction. */
.earthcell-content .ec-c3 .ec-trio__item {
	border-top-width: 2px;
	border-top-color: var(--ec-lime);
}

.earthcell-content .ec-trio__title {
	margin: 0;
	max-width: none;
	font-size: clamp(19px, 2.6vw, 22px);
	line-height: 1.25;
	font-weight: 700;
	letter-spacing: -0.01em;
}

.earthcell-content .ec-trio__item p {
	margin: 10px 0 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.6;
}

/* --------------------------------------------------------------------------
   The three Kogi sites: Munekan Masha, Esmeralda, Duanama.

   THEY ARE NOT THREE EQUAL COLUMNS AND MUST NOT BECOME THEM. The three entries are three, four and
   four paragraphs long, and side by side the shortest leaves a hole under it while the longest
   runs to a thread. They read down the page instead, each under its own name on a left rule, which
   is the shape the design uses for its four-step spine.

   340px on the name column is the widest a name of this length needs. Under about 700px the name
   sits above its own paragraphs instead of beside them, which is what auto-fit does on its own.
   -------------------------------------------------------------------------- */

/*
 * ONE COLUMN PER SITE, NOT A NAME IN ONE COLUMN AND ITS BODY IN ANOTHER.
 *
 * The two-column shape left roughly 350 pixels of empty column beside each of the three sites, and
 * about 1,350 pixels of dead layout in the section a funder is most likely to read. Piers reported
 * it on 29 August 2026 as text chucked in and not unified with the design, and the design review
 * of the same day counted the empty space. The three sites are the product; they should not be the
 * emptiest part of the page.
 *
 * THE MEASURE IS THE ONE THE REST OF THE PAGE USES, so a site reads at the same width as every
 * other passage of prose on the page.
 */
.earthcell-content .ec-sites--stacked .ec-site {
	display: block;
	max-width: var(--ec-measure);
}

.earthcell-content .ec-sites--stacked .ec-site__body {
	max-width: var(--ec-measure);
	margin-top: 14px;
}

.earthcell-content .ec-sites {
	margin-top: clamp(32px, 4.5vw, 52px);
}

.earthcell-content .ec-site {
	display: grid;
	grid-template-columns: repeat(auto-fit, minmax(min(100%, 340px), 1fr));
	gap: clamp(16px, 3vw, 40px);
	padding: clamp(26px, 3.5vw, 38px) 0;
	border-top: 1px solid rgba(62, 68, 84, 0.18);
}

/*
 * THE LAST SITE DROPS ITS RULE. It used to keep one, 88px above the top rule of the next section,
 * which drew two lines with nothing between them and read as an empty section. Found in the
 * design review of 29 August 2026.
 */
.earthcell-content .ec-site:last-child {
	border-bottom: 0;
}

.earthcell-content .ec-site__name {
	margin: 0;
	max-width: none;
	font-size: clamp(22px, 3.4vw, 30px);
	line-height: 1.15;
	font-weight: 800;
	letter-spacing: -0.02em;
}

.earthcell-content .ec-site p {
	margin: 0;
	max-width: var(--ec-measure);
	font-size: 17px;
	line-height: 1.65;
}

/*
 * THE PARAGRAPHS ARE WRAPPED IN ONE BOX, AND THE BOX IS THE GRID CELL.
 *
 * Without it every paragraph was a grid item of its own, so a four-paragraph site laid itself out
 * as the name, then paragraph one beside it, then two under the name, then three beside two: the
 * reader's eye went down, across, down, across, and the entry could not be read at all. A grid
 * lays out its children, and the children have to be the two columns rather than everything in
 * them.
 */
.earthcell-content .ec-site__body {
	max-width: var(--ec-measure);
}

/*
 * The stack gap between those paragraphs. Named tags rather than a universal selector, for the
 * reason written at the top of the .ec-col rules: a bare star loses to the tag that zeroes the
 * margin above it.
 */
.earthcell-content .ec-site__body p + p {
	margin-top: 16px;
}

/* --------------------------------------------------------------------------
   PROJECT PARTNERS. Built 30 August 2026; the chapter was missing entirely.

   IT IS A CONTINUATION, NOT A NEW SUBJECT. The programme chapter above names these organisations
   in passing and this is where they are credited. So it carries no rule above it and none of the
   88px that separates unrelated sections: it opens straight out of the chapter above.

   IT TAKES ITS GROUND FROM THE PAGE, like the other three light chapters. Nothing here sets a
   background except the three plates, which are white on purpose.

   THE BAND IS THE PAGE'S, NOT THIS CHAPTER'S. It is --ec-band, declared once at the top of this
   file. Widening one chapter on its own would step out of line with the programme chapter directly
   above it, which is the one thing this chapter must not do. Whoever re-cuts the band re-cuts it
   for all nine at once, which is what the 4 September 2026 change to 1040px did.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-partners {
	padding-top: clamp(56px, 9vw, 88px);
}

/*
 * ONE COLUMN UNTIL 1160px, THEN THREE, AND NOTHING IN BETWEEN.
 *
 * Three cards in a 900px band give about 265px each, which is roughly 33 characters a line, and a
 * partner's description breaks into a ragged stack. The design flips these rows at 1160px for that
 * reason and holds them at the prose measure below it.
 *
 * IT IS AN EXPLICIT FLIP AND NOT auto-fit. The handoff is emphatic: a per-component auto-fit
 * minimum is what produced two-cards-then-one orphan rows between 760 and 940px elsewhere on this
 * page. Three cards or one card, never two.
 *
 * The design does the flip in script rather than in a media query. A query is used here because the
 * outcome for this row is identical and a stylesheet is where somebody looks for it.
 */
.earthcell-content .ec-partners__row {
	display: grid;
	grid-template-columns: minmax(0, 1fr);
	gap: clamp(20px, 3vw, 32px);
	margin-top: clamp(28px, 4vw, 40px);
	margin-bottom: clamp(28px, 4vw, 40px);
	max-width: var(--ec-measure);
}

@media (min-width: 1160px) {
	.earthcell-content .ec-partners__row {
		grid-template-columns: repeat(3, minmax(0, 1fr));
		max-width: none;
	}
}

.earthcell-content .ec-partner {
	background: #ffffff;
	border: 1px solid rgba(62, 68, 84, 0.18);
	padding: 0;
}

/*
 * A FIXED 124px PLATE, so three marks of three different shapes sit on one line rather than each
 * card being as tall as the picture it was handed. The files supplied are 576x347, 1000x612 and
 * 375x375, and left to themselves the square one would make its card half as tall again.
 *
 * The logo is left-aligned, vertically centred, capped at 76px and contained rather than covered.
 * A partner's mark is never stretched, cropped, or scaled up to fill the space.
 */
.earthcell-content .ec-partner__plate {
	display: flex;
	align-items: center;
	height: 124px;
	padding: 18px clamp(18px, 2.6vw, 26px);
	background: #ffffff;
	border-bottom: 1px solid rgba(62, 68, 84, 0.12);
}

.earthcell-content .ec-partner__logo {
	width: auto;
	max-width: min(100%, 260px);
	max-height: 76px;
	object-fit: contain;
}

/*
 * THE DREAM CHANGE FILE IS THEIR REVERSED MARK: white, drawn for a dark ground. On the white plate
 * the client asked for, it is invisible. Inverting it renders it as black.
 *
 * THIS IS A WORKAROUND AND IT SHOULD NOT SURVIVE. Ask Dream Change for their standard dark or
 * full-colour file and delete this rule the day it arrives. It is safe only because that file is a
 * single flat colour with no detail to damage. Never put this class on a colour logo.
 */
.earthcell-content .ec-partner__logo--invert {
	filter: invert(1);
}

.earthcell-content .ec-partner__body {
	margin: 0;
	padding: clamp(18px, 2.6vw, 26px);
	font-size: clamp(15px, 2vw, 16px);
	line-height: 1.6;
}

.earthcell-content .ec-partner__body a {
	color: var(--ec-teal);
}

/* --------------------------------------------------------------------------
   The four things an EarthCell connects.

   A plain list, and it stays a list. The theme draws a lime circle-arrow on every list item on the
   site, which is switched off here the same way it is for the other two lists on this page. The
   marker is a short teal rule instead, because these four are parts of one sentence that began
   above them rather than four separate claims.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-connects {
	margin: 18px 0 0;
	padding: 0;
	max-width: var(--ec-measure);
	list-style: none;
}

.earthcell-content .ec-connects li::before {
	content: none;
	display: none;
}

.earthcell-content .ec-connects li {
	position: relative;
	margin: 0;
	padding: 7px 0 7px 26px;
	font-size: 17px;
	line-height: 1.6;
}

.earthcell-content .ec-connects li::after {
	content: "";
	position: absolute;
	top: 1.05em;
	left: 0;
	width: 14px;
	height: 1.5px;
	background: var(--ec-teal);
}

/* ==========================================================================
   THE ENQUIRY BAND. Section 09 of the 4 September 2026 design, built 4 September 2026.

   WHY IT IS A BAND AND NOT ANOTHER PASSAGE OF PROSE. This is the only place on the page where a
   visitor can reach C Level, and until today there was no way to reach C Level from this page at
   all: the writing sent the reader up to the map and the map sent them back down to the writing.
   Now the form exists, and it has to stop looking like the end of an article. A full width change
   of ground is the cheapest way to say the reading has finished and the asking has started.

   THE FORM ITSELF IS DRAWN BY THE WIDGET, not by the page and not by this file. The page supplies
   an empty div carrying `data-ec-enquiry-form` and the widget renders into it, which is what lets
   the form name the EarthCells the visitor picked on the map above. This file paints the ground the
   form stands on, and the card it stands in, and nothing else. If the widget is not loaded the
   band is still correct: a teal band with three paragraphs on it and no card.

   THE FULL BLEED SUM IS THE ONE .ec-full AND .ec-band ALREADY USE, and it is not a plain 100vw.
   100vw includes the vertical scrollbar and this page always has one, so a plain 100vw band is
   about 15px too wide and shifts left, which was measured on a 1536px window on 28 August 2026 and
   cut the imagery credit off the map. --ec-scrollbar is that difference and it is 0 on a phone.
   ========================================================================== */

/*
 * WHY THIS BAND DOES NOT USE --ec-scrollbar, WHICH EVERY OTHER FULL WIDTH THING HERE USES.
 *
 * That token holds the text `calc(100vw - 100%)`, and a custom property is substituted where it is
 * USED, not where it is declared. So its `100%` means the width of whatever element reads it. On
 * `.ec-full` that is the article, which is the width of the window, and the sum really is the
 * scrollbar. This band sits inside the 1040px reading column instead, so its `100%` is 1040px and
 * the same token claims the scrollbar is about 400 pixels wide. Measured on staging, 4 September
 * 2026: the band came out 992px inside a 1425px page and reached neither edge.
 *
 * SO THIS ONE IS A PLAIN 100vw AND OVERHANGS BY THE SCROLLBAR, about 15px, split between the two
 * sides. `.earthcell-article` already carries `overflow-x: clip`, which swallows the overhang, and
 * the page does not scroll sideways. That was checked rather than assumed.
 *
 * CLIP IS THE WRONG ANSWER WHEN SOMETHING IS ANCHORED TO THE EDGE, and it was rejected for exactly
 * that reason on the map above, where it hid the zoom buttons and the imagery credit the licence
 * requires be shown. Nothing here is anchored to the band's own edge. The card and the prose are
 * placed by the padding below, which centres them on the same 1040px column as the rest of the
 * page, a long way inside the clipped strip.
 *
 * A REGISTERED PROPERTY WAS TRIED FIRST AND DOES NOT WORK. `@property` with a `<length>` syntax
 * would compute at `:root`, where `100%` finally means the root's own width. But a percentage is
 * not a valid computed `<length>`, so the declaration is dropped and the token silently falls back
 * to its initial value. The band was the right width for the wrong reason for about a minute.
 */
.earthcell-content > #enquiry {
	width: 100vw;
	margin-left: calc(50% - 50vw);
	margin-top: clamp(64px, 11vw, 104px);
	/*
	 * THE GROUND RUNS TO BOTH EDGES AND THE WRITING DOES NOT. The band is the full window; its
	 * contents stay in the same 1040px column as everything above, so the enquiry lines up with the
	 * prose it follows rather than sprawling across a wide screen. The side padding is whatever it
	 * takes to centre that column, and never less than the 24px the rest of the page uses.
	 */
	padding-block: clamp(56px, 9vw, 96px);
	padding-inline: max(24px, calc((100vw - var(--ec-band)) / 2));
	background: var(--ec-teal);
	color: var(--ec-paper);
	/*
	 * TWO COLUMNS, AND EVERY CHILD IS PLACED RATHER THAN LEFT TO FLOW.
	 *
	 * `auto-fit` alone was tried first and was wrong here, because this section has five children
	 * in one flat list: a heading, three paragraphs and the form. Left to flow they filled the grid
	 * in reading order, so the second paragraph landed to the right of the first, the third went
	 * back to the left, and the form ended up under a column of white space. Seen on staging on
	 * 4 September 2026.
	 *
	 * The page body cannot be changed to wrap the prose in a div of its own, because it is
	 * WordPress content and a builder save would strip it. So the placement lives here instead.
	 */
	display: grid;
	grid-template-columns: 1fr minmax(330px, 460px);
	align-items: start;
	/*
	 * THE GAP IS BETWEEN THE COLUMNS ONLY, AND THERE IS A HARD REASON FOR IT.
	 *
	 * The card spans a generous number of rows so it can stand beside however many paragraphs the
	 * copy has. A row gap applies between every one of those rows, including the empty ones past
	 * the end of the prose, so a plain `gap` added about 5,500 pixels of nothing and the section
	 * came out 6,172 pixels tall. Measured on staging, 4 September 2026.
	 *
	 * The paragraphs carry their own spacing, which is the ordinary prose rhythm and is what the
	 * rest of the page uses anyway.
	 */
	column-gap: clamp(32px, 5vw, 56px);
	row-gap: 0;
	max-width: none;
}

/*
 * WHEN THE SECTION'S CONTENT IS WRAPPED, THE TWO COLUMNS ARE WRONG AND MUST NOT APPLY.
 *
 * Found by Piers on 6 September 2026, on the Kogi page: "the most crucial bit of the whole page
 * isn't working." The EarthCell picking map was rendering 116 pixels wide, with the imagery credit
 * wrapping vertically over it in a white block.
 *
 * THE CAUSE. The grid above was written on 4 September 2026 for a section whose children were a
 * flat list: a heading, three paragraphs and the form. The Kogi page's markup wraps all of that in
 * a single `.ec-c8__inner` instead, so the grid has ONE item. That item took the 1fr column, the
 * `minmax(330px, 460px)` column stayed empty, and everything inside was squeezed into 524 pixels
 * of a 984 pixel band. `.ec-choose` then split those 524 into its own map-plus-panel pair and the
 * map got what was left, which was 116.
 *
 * IT WAS NOT THE DESIGN PASS. Switching css/earthcell-tokens.css off in the browser leaves the map
 * at 116 pixels, so this has been the case since the wrapper and the grid met. The Bioestrela pages
 * have no wrapper, their map is 632 pixels, and that is what this one should look like.
 *
 * A single wrapped child spans both columns. The wrapper is a flex column and does its own layout,
 * which is what it was written to do.
 */
.earthcell-content > #enquiry > .ec-c8__inner {
	grid-column: 1 / -1;
	max-width: none;
	width: 100%;
}

/*
 * THE HEADING AND THE PROSE SHARE THE LEFT COLUMN. The heading spans both columns, because a
 * heading that sits in one half of a two column band reads as a label for that half only.
 */
.earthcell-content > #enquiry > h2 {
	grid-column: 1 / -1;
	max-width: var(--ec-heading-measure, min(600px, 100%));
	margin: 0;
	color: var(--ec-paper);
}

/* Every paragraph stays in the left column, under the heading, in the order it was written. */
.earthcell-content > #enquiry > p {
	grid-column: 1;
	max-width: 46ch;
	color: var(--ec-paper);
}

/*
 * The card sits in the right column, starting level with the first paragraph.
 *
 * IT HAS TO SPAN EVERY ROW THE PROSE USES, and `span 99` is how, because the number of rows is
 * whatever WordPress made: the section holds four paragraphs, one of which is empty, and that count
 * changes the day anybody edits the copy. Pinned to row 2 alone the card made row 2 as tall as
 * itself, and the remaining paragraphs were pushed 700 pixels down the page with nothing beside
 * them. Seen on staging on 4 September 2026.
 *
 * The rows past the end of the prose are empty, so they take `grid-auto-rows: min-content` and
 * collapse to no height at all. The section ends up as tall as the taller of the two columns, which
 * is what it should have been all along.
 */
.earthcell-content > #enquiry {
	grid-auto-rows: min-content;
}

.earthcell-content > #enquiry > .ec-enquiry {
	grid-column: 2;
	grid-row: 2 / span 99;
	align-self: start;
}

/*
 * The last paragraph is the one that says nothing here is a commitment. It is quieter than the two
 * above it, and the drop in size does the work rather than a drop in contrast: dimming small text
 * with opacity is how the earlier iteration failed the contrast standard at 3.3 to 1.
 */
.earthcell-content > #enquiry > p:last-of-type {
	font-size: 0.94rem;
}

/*
 * THE CARD. Paper on teal, which the token audit measures at 7.4 to 1, so the slate text inside it
 * sits on the ground it was designed for rather than on the band.
 */
.earthcell-content > #enquiry > .ec-enquiry {
	background: var(--ec-paper);
	color: var(--ec-ink);
	border-radius: 10px;
	padding: clamp(18px, 4vw, 28px);
	max-width: 460px;
	width: 100%;
	justify-self: end;
}

/*
 * ON ONE COLUMN THE CARD STOPS HUGGING THE RIGHT EDGE. `justify-self: end` is right beside the
 * prose and wrong underneath it, where it would sit against the band's right padding with a ragged
 * gap on the left. Keyed off the grid having collapsed, not off a device width.
 */
/*
 * ONE COLUMN ONCE THE PAIR STOPS FITTING. The two columns need about 330px each plus the gap, so
 * below roughly 760px the form goes under the prose and every placement above is handed back to the
 * normal flow. The card also stops hugging the right edge, which is right beside the prose and
 * wrong underneath it, where it would sit against the padding with a ragged gap on the left.
 */
@media (max-width: 760px) {
	.earthcell-content > #enquiry {
		grid-template-columns: 1fr;
	}

	.earthcell-content > #enquiry > p,
	.earthcell-content > #enquiry > .ec-enquiry {
		grid-column: auto;
		grid-row: auto;
	}

	.earthcell-content > #enquiry > .ec-enquiry {
		justify-self: stretch;
		max-width: none;
	}
}


/* ==========================================================================
   THE TONAL BAND SYSTEM.

   From the handoff "EarthCell page section rhythm", 6 September 2026, direction
   1a. Piers: the page "reads as one continuous scroll: its sections sit on the
   same background with, at most, a hairline rule between them, so a reader
   cannot tell where one part ends and the next begins."

   EVERY SECTION GETS ITS OWN GROUND AND NO TWO NEIGHBOURS SHARE ONE. The change
   of colour IS the separator. That is why there are no rules between bands: a
   hairline and a colour change are two separators doing one job, and the
   hairline is the weaker of the two.

   THE SEQUENCE, and adjacency is the whole point of it. If a section is ever
   added, removed or reordered, check this column first:

     hero                night   #05080B   the map instrument, unchanged
     01 .ec-c4           paper   #F4F8F9   was black
     02 .ec-c3           teal    #005A69
     03 .ec-section--rule + .ec-partners   paper   one band, one topic
     04 .ec-c5           tint    #E7EEF0
     05 .ec-c6           teal    #005A69
     06 .ec-section--narrow     paper
     07 .ec-c8           lime    #CEF17B
     -- .ec-c10          tint    #E7EEF0   a footnote, not a section

   THIS REPLACES THE ONE-GROUND VERSION of earlier the same day, which put every
   chapter on paper with a hairline between. That fixed the black voids and left
   the page with nothing to divide it, which is the fault this handoff answers.
   The superseded rules were removed from css/earthcell-tokens.css rather than
   left to fight these; see the note in that file.

   CONTRAST, MEASURED AND ALL PASSING AA FOR BODY TEXT: slate on paper 9.1:1,
   slate on tint 8.6:1, paper on teal 7.4:1, slate on lime 7.6:1, paper on night
   15.8:1. Never white on lime, which is 1.2:1.

   ONE DEVIATION FROM THE HANDOFF, AND THE EVIDENCE FOR IT. The handoff keeps
   .ec-c3 and .ec-c6 as two-column layouts with a chapter map on one side. Those
   map containers have never drawn: tested on 6 September 2026 with the widget
   loaded, the containers un-hidden at 360px and eight seconds to fill, all five
   stayed empty, no canvas and no children, on staging and on live alike. A
   two-column band with an empty column is a coloured rectangle beside the
   words, which is the fault the handoff warns about in its own note on
   data-ec-chapter. So those bands run one column until the widget fills them.
   The markup and the data-ec-chapter-map attributes are untouched, so the day
   it does fill them, deleting the two rules marked ONE COLUMN UNTIL THE MAP
   DRAWS restores the handoff exactly.
   ========================================================================== */

.earthcell-content {

	/* The handoff's tokens. Night, tint and the greys are derived; the rest are
	   the theme's own. --theme-pink and --theme-pinker are deliberately absent:
	   neither carries body text at an accessible ratio on any of these grounds. */
	--ec-night: #05080b;
	--ec-tint: #e7eef0;
	--ec-lime: #cef17b;
	--ec-reversed-body: #e2eef0;
	--ec-reversed-credit: #a9cccf;
	--ec-muted-caption: #5d6672;
	--ec-muted-year: #6d7583;
	--ec-eyebrow-on-lime: #28513f;

	/* 72 top and bottom is one of the handoff's two deliberate exceptions to the
	   4 / 8 / 12 / 16 / 24 / 32 / 48 / 64 / 96 scale. The other is the 44px grid
	   gap. Both are named here so neither looks like a stray number. */
	--ec-band-pad: 72px;
	--ec-band-gap: 44px;
}

@media ( max-width: 900px ) {
	.earthcell-content { --ec-band-pad: 56px; }
}

@media ( max-width: 600px ) {
	.earthcell-content { --ec-band-pad: 40px; }
}

/* --------------------------------------------------------------------------
   EVERY BAND IS FULL BLEED, WITH THE CONTENT COLUMN INSIDE IT.

   The ground runs to both edges of the window; the words stay in the container
   the page already uses. --ec-scrollbar is the file's own calc(100vw - 100%)
   and is why this does not scroll sideways; see the note beside .ec-band.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c4,
.earthcell-content > .ec-c3,
.earthcell-content > .ec-section--rule,
.earthcell-content > .ec-partners,
.earthcell-content > .ec-c5,
.earthcell-content > .ec-c6,
.earthcell-content > .ec-section--narrow,
.earthcell-content > .ec-c8,
.earthcell-content > .ec-c10 {
	width: 100vw;
	/*
	 * THE FILE'S OWN FULL-BLEED SUM, USED SIX TIMES ABOVE THIS LINE, AND IT IS
	 * USED HERE FOR THE SAME REASON RATHER THAN A CLEVERER ONE.
	 *
	 * The first version of this rule was
	 *   calc( 50% - 50vw + ( var(--ec-scrollbar) / 2 ) )
	 * meaning to centre on the visible window rather than the layout viewport.
	 * --ec-scrollbar is itself a calc, so the substitution nests one calc inside
	 * a division inside another, and the browser threw the whole declaration
	 * away. An invalid calc does not fall back to something sensible: the
	 * property reverts to 0, so every band kept its container's left edge and
	 * sat 456px to the right of where it should on a 1920px window, with the
	 * photograph and its caption running off the screen. Piers saw it before I
	 * did, because I had measured at 1440 where the number is smaller.
	 *
	 * A silently dropped declaration is the third failure of this shape today.
	 * If a rule seems not to apply, check whether it is valid before checking
	 * specificity: .earthcell-article already carries overflow-x: clip, which
	 * is what absorbs the half-scrollbar this sum does not subtract.
	 */
	margin-left: calc( 50% - 50vw );
	margin-right: 0;
	margin-top: 0;
	margin-bottom: 0;
	padding-top: var(--ec-band-pad);
	padding-bottom: var(--ec-band-pad);
	padding-left: var(--ec-gutter);
	padding-right: var(--ec-gutter);
}

/* The inner column of every band holds the page's own width. */

.earthcell-content > .ec-c4 > *,
.earthcell-content > .ec-c3 > *,
.earthcell-content > .ec-section--rule > *,
.earthcell-content > .ec-partners > *,
.earthcell-content > .ec-c5 > *,
.earthcell-content > .ec-c6 > *,
.earthcell-content > .ec-section--narrow > *,
.earthcell-content > .ec-c8 > *,
.earthcell-content > .ec-c10 > * {
	max-width: var(--ec-band);
	margin-left: auto;
	margin-right: auto;
	padding-top: 0;
	padding-bottom: 0;
}

/*
 * 03 IS TWO SECTIONS SHARING ONE GROUND, WHICH IS CORRECT: they are one topic.
 * So the programme section pays the top padding, the partners section the
 * bottom, and neither pays in the middle. Without this the reader gets 144px of
 * paper in the middle of one band and reads it as two.
 */
.earthcell-content > .ec-section--rule { padding-bottom: 0; }
.earthcell-content > .ec-partners { padding-top: 0; }

/* NO HAIRLINES BETWEEN BANDS. The change of ground is the separator. */

.earthcell-content > .ec-section--rule,
.earthcell-content > .ec-partners,
.earthcell-content > .ec-c4,
.earthcell-content > .ec-c3,
.earthcell-content > .ec-c5,
.earthcell-content > .ec-c6,
.earthcell-content > .ec-c8,
.earthcell-content > .ec-c10 {
	border-top: 0;
	border-bottom: 0;
}

/* --------------------------------------------------------------------------
   THE GROUNDS.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c4,
.earthcell-content > .ec-section--rule,
.earthcell-content > .ec-partners,
.earthcell-content > .ec-section--narrow {
	background: var(--ec-paper);
	color: var(--ec-ink);
}

.earthcell-content > .ec-c3,
.earthcell-content > .ec-c6 {
	background: var(--ec-teal);
	color: var(--ec-paper);
}

.earthcell-content > .ec-c5,
.earthcell-content > .ec-c10 {
	background: var(--ec-tint);
	color: var(--ec-ink);
}

.earthcell-content > .ec-c8 {
	background: var(--ec-lime);
	color: var(--ec-ink);
}

/* Text on the light bands. */

.earthcell-content > .ec-c4 :is( h2, h3, p, li, .ec-c4__title, .ec-c4__body ),
.earthcell-content > .ec-section--rule :is( h2, h3, p, li ),
.earthcell-content > .ec-partners :is( h2, h3, p, li ),
.earthcell-content > .ec-section--narrow :is( h2, h3, p, li ),
.earthcell-content > .ec-c5 :is( h2, h3, p, li, .ec-c5__title, .ec-c5__body ),
.earthcell-content > .ec-c8 :is( h2, h3, p, li, label, legend ) {
	color: var(--ec-ink);
}

/* Text on the teal bands. Body copy takes the reversed tone, not pure paper. */

.earthcell-content > .ec-c3 :is( h2, h3, .ec-c3__title, .ec-trio__title ),
.earthcell-content > .ec-c6 :is( h2, h3, .ec-c6__title ) {
	color: var(--ec-paper);
}

.earthcell-content > .ec-c3 :is( p, li, .ec-c3__body ),
.earthcell-content > .ec-c6 :is( p, li ) {
	color: var(--ec-reversed-body);
}

.earthcell-content > .ec-c3 a:not( .ec-btn ),
.earthcell-content > .ec-c6 a:not( .ec-btn ) {
	color: var(--ec-lime);
}

/* --------------------------------------------------------------------------
   MONO LABELS. One treatment for every eyebrow, caption, credit and status
   line, coloured by the ground it sits on.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-eyebrow,
.earthcell-content .ec-credit,
.earthcell-content .ec-c4__note,
.earthcell-content .ec-c6__label,
.earthcell-content .ec-c6__caption,
.earthcell-content .ec-c5__status,
.earthcell-content .ec-c5__year,
.earthcell-content .ec-c8__label,
.earthcell-content .ec-marker {
	font-family: var(--ec-mono);
	font-weight: 600;
	line-height: 1;
	letter-spacing: 0.16em;
	text-transform: uppercase;
}

.earthcell-content .ec-eyebrow {
	display: block;
	margin: 0 0 20px;
	font-size: 11px;
	color: var(--ec-teal);
}

.earthcell-content .ec-credit,
.earthcell-content .ec-c4__note {
	font-size: 10px;
	line-height: 1.6;
	letter-spacing: 0.1em;
	font-weight: 600;
	color: var(--ec-muted-caption);
}

/* Eyebrows and credits change colour with the ground, and only the colour. */

.earthcell-content > .ec-c3 .ec-eyebrow,
.earthcell-content > .ec-c6 .ec-eyebrow {
	color: var(--ec-lime);
}

.earthcell-content > .ec-c3 .ec-credit,
.earthcell-content > .ec-c6 .ec-credit,
.earthcell-content > .ec-c3 .ec-c6__caption,
.earthcell-content > .ec-c6 .ec-c6__caption {
	color: var(--ec-reversed-credit);
}

/*
 * THE ONE COLOUR OUTSIDE THE TOKEN SET, AND IT IS DELIBERATE. #005A69 on lime
 * is too close in value to hold an 11px mono label. This is the only place it
 * appears and it must not spread.
 */
.earthcell-content > .ec-c8 .ec-eyebrow {
	color: var(--ec-eyebrow-on-lime);
}

/* --------------------------------------------------------------------------
   01 .ec-c4 THE HEART OF THE WORLD, ON PAPER.
   The strongest break on the page: a reader leaves a black instrument and
   arrives on paper. The photograph stays a contained figure, never a ground.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c4 {
	background-image: none;
}

.earthcell-content .ec-c4__inner {
	display: grid;
	grid-template-columns: 1.15fr 0.85fr;
	gap: var(--ec-band-gap);
	align-items: start;
}

.earthcell-content .ec-c4__title {
	font-size: 40px;
	line-height: 1.12;
	font-weight: 800;
	max-width: 20ch;
}

.earthcell-content .ec-c4__body {
	font-size: 18px;
	line-height: 1.62;
	max-width: 58ch;
}

.earthcell-content .ec-c4__words > * + .ec-c4__body {
	margin-top: 16px;
}

.earthcell-content .ec-c4__figure::after {
	content: none;
}

.earthcell-content .ec-c4__note--photo {
	margin-top: 10px;
}

/* --------------------------------------------------------------------------
   02 .ec-c3 THREE SIMPLE PRINCIPLES, ON TEAL.
   -------------------------------------------------------------------------- */

/*
 * ONE COLUMN UNTIL THE MAP DRAWS. See the note at the top of this block for the
 * test. Delete this rule and the one on .ec-c6 to restore the handoff's two
 * columns the day the widget fills data-ec-chapter-map.
 */
.earthcell-content .ec-c3__inner {
	display: grid;
	grid-template-columns: minmax( 0, 1fr );
	gap: var(--ec-band-gap);
	align-items: start;
}

.earthcell-content .ec-c3__map {
	display: none;
}

.earthcell-content .ec-c3__words {
	padding-top: 0;
	padding-bottom: 0;
}

.earthcell-content .ec-c3__title {
	font-size: 40px;
	line-height: 1.12;
	font-weight: 800;
	max-width: 20ch;
}

.earthcell-content .ec-c3__body {
	font-size: 18px;
	line-height: 1.62;
	max-width: 62ch;
}

.earthcell-content .ec-trio {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 22px;
	margin-top: 32px;
}

.earthcell-content .ec-trio__item {
	border-top: 2px solid var(--ec-aqua);
	padding-top: 12px;
}

.earthcell-content .ec-trio__title {
	font-size: 19px;
	font-weight: 700;
	margin-bottom: 6px;
	color: var(--ec-paper);
}

.earthcell-content .ec-trio__item p {
	font-size: 16px;
	line-height: 1.6;
	color: var(--ec-reversed-body);
}

/* --------------------------------------------------------------------------
   03 THE PROGRAMME AND THE PARTNERS, ON PAPER.
   The one internal rule left on the page divides two parts of one band.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-section--rule h2,
.earthcell-content > .ec-section--narrow h2,
.earthcell-content > .ec-c5 h2,
.earthcell-content > .ec-c8 h2 {
	font-size: 40px;
	line-height: 1.12;
	font-weight: 800;
	max-width: 22ch;
}

.earthcell-content > .ec-partners {
	margin-top: 0;
}

.earthcell-content > .ec-partners > * {
	border-top: 1px solid rgba( 62, 68, 84, 0.18 );
	padding-top: 32px;
	margin-top: 48px;
}

/* --------------------------------------------------------------------------
   04 .ec-c5 FOLLOW WHAT HAPPENS NEXT, ON TINT.
   The one band whose colour exists only to break two paper sections apart. It
   must stay a tint of paper and never become a fifth colour.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c5__title {
	font-size: 40px;
	line-height: 1.12;
	font-weight: 800;
	max-width: 22ch;
}

.earthcell-content .ec-c5__body {
	font-size: 18px;
	line-height: 1.62;
	max-width: 62ch;
}

.earthcell-content .ec-c5__map {
	display: none;
}

.earthcell-content .ec-c5__year {
	font-size: 13px;
	font-weight: 700;
	color: var(--ec-ink);
}

.earthcell-content .ec-c5__year:not( :first-child ) {
	color: var(--ec-muted-year);
}

.earthcell-content .ec-c5__status {
	font-size: 10px;
	letter-spacing: 0.1em;
	color: var(--ec-muted-caption);
}

.earthcell-content .ec-c5__status--real {
	color: var(--ec-teal);
}

/* --------------------------------------------------------------------------
   05 .ec-c6 THREE CAPTURE SCALES, ON TEAL.
   The panels sit inside the container on teal rather than meeting the band
   edges, which is what stops this reading as a continuation of the hero.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c6__row {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 24px;
}

/* ONE COLUMN UNTIL THE MAP DRAWS. See .ec-c3__inner above. */
.earthcell-content .ec-c6__map {
	display: none;
}

.earthcell-content .ec-c6__panel {
	position: relative;
	min-height: 0;
	padding: 0;
	background: none;
	border: 0;
	overflow: visible;
}

.earthcell-content .ec-c6__panel > * {
	position: static;
	inset: auto;
}

.earthcell-content .ec-c6__label {
	display: block;
	margin: 0 0 6px;
	font-size: 12px;
	font-weight: 700;
	letter-spacing: 0.16em;
	color: var(--ec-lime);
	text-shadow: none;
	background: none;
}

.earthcell-content .ec-c6__caption {
	font-size: 16px;
	line-height: 1.55;
	font-family: inherit;
	font-weight: 400;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-reversed-body);
	text-shadow: none;
	background: none;
}

/* --------------------------------------------------------------------------
   06 .ec-section--narrow WHAT IS AN EARTHCELL?, ON PAPER.
   One step larger than the rest of the page, because this is the definition
   the whole page turns on.

   IT NAMES THE LEDE NOW, NOT WHATEVER COMES FIRST. This was
   `> p:first-of-type`, written when the section opened with its lede. An
   eyebrow was added above the heading on 6 September 2026, which made the
   eyebrow the first paragraph. So the eyebrow took 19px and the lede lost the
   size the rule existed to give it. Nothing looked broken; two elements were
   just each wearing the other's size.

   `h2 + p` is the lede by position and cannot be taken by anything added above
   the heading. See the note on the eyebrow rule at the foot of this file for
   the other half of the same fault.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-section--narrow > * > h2 + p,
.earthcell-content > .ec-section--narrow > h2 + p {
	font-size: 20px;
	line-height: 1.6;
	max-width: 62ch;
}

/* --------------------------------------------------------------------------
   07 .ec-c8 HELP FUND THE NEXT STAGE, ON LIME.
   NEVER WHITE TEXT ON LIME: the ratio is 1.2:1. Slate only.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c8 .ec-c8__label,
.earthcell-content > .ec-c8 .ec-marker,
.earthcell-content > .ec-c8 .ec-c8__chip-status {
	color: var(--ec-eyebrow-on-lime);
	text-shadow: none;
}

.earthcell-content > .ec-c8 .ec-marker--demo {
	color: #4a5a52;
}

.earthcell-content > .ec-c8 :is( input, textarea, select ) {
	background: var(--ec-paper);
	border: 1px solid rgba( 62, 68, 84, 0.3 );
	color: var(--ec-ink);
	border-radius: 0;
}

.earthcell-content > .ec-c8 .ec-btn,
.earthcell-content > .ec-c8 button[type="submit"] {
	background: var(--ec-teal);
	color: var(--ec-paper);
	font-family: var(--ec-mono);
	font-size: 13px;
	font-weight: 700;
	letter-spacing: 0.12em;
	padding: 16px 26px;
	border-radius: 0;
	border: 0;
}

/*
 * THE PICKING MAP IS THE ONE DARK THING LEFT IN THIS BAND, and it stays dark
 * because it is satellite imagery. Its panel keeps its own colours so the
 * lime ground never reaches inside the instrument.
 */
.earthcell-content > .ec-c8 .ec-choose__panel {
	color: var(--ec-paper);
}

.earthcell-content > .ec-c8 .ec-choose__panel :is( p, li, h2, h3 ) {
	color: inherit;
}

/* --------------------------------------------------------------------------
   THE PROTOTYPE NOTE, ON TINT. A footnote to the page, not a section, which is
   why it takes the tint and not paper: it must not read as another band.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c10 {
	padding-top: 28px;
	padding-bottom: 28px;
}

.earthcell-content > .ec-c10,
.earthcell-content > .ec-c10 p {
	font-family: var(--ec-mono);
	font-size: 10px;
	line-height: 1.7;
	color: var(--ec-muted-caption);
}

/* --------------------------------------------------------------------------
   RESPONSIVE. The grounds never change with the viewport: on a phone, where a
   band may be taller than the screen, the colour change is the only cue a
   reader gets, so it has to survive at every width.
   -------------------------------------------------------------------------- */

@media ( max-width: 900px ) {

	.earthcell-content .ec-c4__inner,
	.earthcell-content .ec-c3__inner,
	.earthcell-content .ec-trio,
	.earthcell-content .ec-c6__row {
		grid-template-columns: minmax( 0, 1fr );
	}
}

@media ( max-width: 600px ) {

	.earthcell-content .ec-c4__title,
	.earthcell-content .ec-c3__title,
	.earthcell-content .ec-c5__title,
	.earthcell-content > .ec-section--rule h2,
	.earthcell-content > .ec-section--narrow h2,
	.earthcell-content > .ec-c5 h2,
	.earthcell-content > .ec-c8 h2 {
		font-size: 32px;
	}

	.earthcell-content .ec-c4__body,
	.earthcell-content .ec-c3__body,
	.earthcell-content .ec-c5__body {
		font-size: 17px;
	}
}

/* --------------------------------------------------------------------------
   TWO CORRECTIONS FOUND BY MEASURING THE BANDS AFTER THEY WENT UP.

   1. THE ASK BAND STAYED TEAL, AND THAT LEFT INVISIBLE TEXT.

   `.earthcell-content > #enquiry` sets the teal ground and paper text. It is an
   ID selector, so it beat every `.ec-c8` rule above: the ground stayed teal
   while the text rules did apply and turned the headings slate. Slate on teal
   measures 1.14:1, which is text you cannot see at all, on the one band that
   carries the ask.

   A class cannot outrank an ID. The rules below name the ID, which is the only
   honest way to win this and is why they sit apart from the band block rather
   than being folded into it.

   2. THE DEFINITION BAND WAS NOT FULL BLEED.

   `.ec-section--narrow` carries its own `max-width`, so `width: 100vw` had
   nothing to do. The cap moves to its children, which is where a reading
   measure belongs anyway.
   -------------------------------------------------------------------------- */

.earthcell-content > #enquiry.ec-c8 {
	background: var(--ec-lime);
	color: var(--ec-ink);
	padding-top: var(--ec-band-pad);
	padding-bottom: var(--ec-band-pad);
}

.earthcell-content > #enquiry.ec-c8 :is( h2, h3, p, li, label, legend ) {
	color: var(--ec-ink);
}

.earthcell-content > #enquiry.ec-c8 .ec-eyebrow {
	color: var(--ec-eyebrow-on-lime);
}

/* The picking map is satellite imagery and stays dark, so its own panel keeps
   its own light-on-dark colours and the lime never reaches inside it. */
.earthcell-content > #enquiry.ec-c8 .ec-choose__panel :is( h2, h3, p, li ) {
	color: var(--ec-paper);
}

.earthcell-content > .ec-section--narrow {
	max-width: none;
}

.earthcell-content > .ec-section--narrow > * {
	max-width: var(--ec-band);
	margin-left: auto;
	margin-right: auto;
}

/* --------------------------------------------------------------------------
   THE BAND COLOUR RULES MUST NOT REACH INSIDE A COMPONENT THAT PAINTS ITS OWN
   SURFACE. This is the third time in one day that a colour was set without
   checking what it would land on, so it is worth naming the pattern.

   "This band is light, therefore its text is ink" is true of prose and false of
   anything sitting on its own dark ground inside that band. Measured after the
   bands went up:

     .ec-sitecard__scale        ink on a satellite thumbnail   2.06:1
     .ec-choose__fallback       ink on the picking map         2.06:1
     .ec-choose__readout        ink on the picking map         1.75:1
     .ec-choose__view-eyebrow   ink on the picking map         2.01:1
     .ec-c8__chip-code          paper on lime                  1.19:1

   Every one of those was legible before the band rules and was broken by them.
   A rule that sets colour by ancestor has to stop at the first descendant that
   sets its own background, and CSS will not do that on its own.
   -------------------------------------------------------------------------- */

/* Labels laid over imagery keep the light tone the imagery needs.

   .ec-sitecard__scale IS NO LONGER ONE OF THEM, AND THIS RULE WAS WHAT KEPT IT
   INVISIBLE. 7 September 2026.

   A later block in this file already said "the label sits under the picture
   now" and set it static, ordered under the thumbnail, and teal. None of that
   reached the page. Two things stopped it. The markup still had the label
   inside .ec-sitecard__diagram, so `order` had nothing to order. And this rule
   is three classes to that one's two, so the paper colour won anyway.

   The result on staging was a line of paper text sitting directly on bright
   satellite imagery. Sampled off the rendered page across the strip the words
   occupy: 1.60:1 at its worst. Nobody could read "Smaller plot · already
   purchased" and nobody had noticed, because an invisible line looks like a
   line that was never there.

   The label is out of the picture in the page body now, and teal on paper is
   what the reference does. Nothing is laid over the thumbnail any more. */

.earthcell-content .ec-sitecard :is( .ec-sitecard__label ) {
	color: var(--ec-paper);
}

/* The picking map is a dark instrument inside a light band. It keeps its own. */

.earthcell-content .ec-choose :is( p, li, h2, h3, span, button, label ),
.earthcell-content .ec-choose__panel :is( p, li, h2, h3 ),
.earthcell-content .ec-choose__fallback,
.earthcell-content .ec-choose__readout,
.earthcell-content .ec-choose__view-eyebrow {
	color: var(--ec-paper);
}

/*
 * THE SELECTED-CELL CHIPS SIT ON LIME NOW. Paper on lime is 1.19:1. Slate is
 * 7.6:1 and is the only text colour this ground takes, which is the handoff's
 * own rule for band 07.
 */
.earthcell-content > #enquiry.ec-c8 .ec-c8__chip,
.earthcell-content > #enquiry.ec-c8 .ec-c8__chip-code,
.earthcell-content > #enquiry.ec-c8 .ec-c8__chip-status {
	color: var(--ec-ink);
}

/*
 * THE PHOTOGRAPH CAPTION AND THE INACTIVE YEARS take the handoff's muted tones.
 * The caption was #8C9BA3, which is 2.68:1 on paper. The handoff says #5d6672,
 * which is 4.9:1. The inactive years measured 3.95:1 on tint at the handoff's
 * own #6d7583, so they take the caption tone instead: it is the same family,
 * one step darker, and it passes on the ground it actually sits on.
 */
.earthcell-content > .ec-c4 .ec-c4__note,
.earthcell-content > .ec-c4 .ec-c4__note--photo {
	color: var(--ec-muted-caption);
}

.earthcell-content > .ec-c5 .ec-c5__year:not( :first-child ) {
	color: var(--ec-muted-caption);
}

/*
 * THE LAST FOUR, AND WHY THE RULE ABOVE MISSED THEM.
 *
 * That rule listed tags - p, li, h2, h3, span, button, label. These four are
 * divs, so it never touched them and they kept the ink the band rule gave
 * them: 2.06:1, 1.75:1 and 2.01:1 on the dark map. Naming tags inside a
 * component is a guess about its markup; naming the component's own classes is
 * not.
 *
 * .ec-choose__picture is the odd one out. It is the view switcher's label and
 * it sits OUTSIDE the map on the lime band, where paper measures 1.19:1, so it
 * takes ink rather than paper.
 */
.earthcell-content .ec-choose__fallback,
.earthcell-content .ec-choose__readout,
.earthcell-content .ec-choose__view-eyebrow {
	color: var(--ec-paper);
}

.earthcell-content > #enquiry.ec-c8 .ec-choose__picture {
	color: var(--ec-ink);
}

/* --------------------------------------------------------------------------
   THE BAND PADDING WAS BEING CANCELLED BY TWO RULES FROM THE OLD DESIGN.

   Piers: "Needs more padding." The eyebrow "03 · THE KOGI PROGRAMME" was
   sitting about ten pixels under the teal band above it, with no breathing
   room at all. It was not that 72px was too little. It was that 72px was never
   applied.

   TWO SURVIVORS OF THE PREVIOUS SYSTEM WERE WINNING:

   1. `.earthcell-content .ec-c3 + .ec-section--rule { padding-top: 0 }` is three
      classes deep and beat the band rule's two. It was written when a dark
      chapter met a light section and the two paddings doubled up. In a band
      system that is exactly the space the reader needs, so it must not fire.

   2. `.earthcell-content > h2` carries `margin: 88px 0 1.5rem`. Inside a band
      the band already supplies the space above a heading, so that 88 was either
      adding to the padding or, where the padding was cancelled, standing in for
      it unevenly. A band owns its own top space; nothing inside it may add to
      it or replace it.

   THE LESSON, AND IT IS THE FOURTH OF THIS SHAPE TODAY: when a rule seems not
   to apply, look for an older rule that is more specific before assuming the
   value is wrong. Every one of these was a correct number defeated by a
   selector nobody remembered.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c1 + .ec-section--rule,
.earthcell-content > .ec-c3 + .ec-section--rule,
.earthcell-content > .ec-c4 + .ec-section--rule,
.earthcell-content > .ec-c5 + .ec-section--rule,
.earthcell-content > .ec-c6 + .ec-section--rule,
.earthcell-content > .ec-section--rule {
	padding-top: var(--ec-band-pad);
	padding-bottom: 0;
	border-top: 0;
	margin-top: 0;
}

/* A heading inside a band takes no top margin: the band is the space. */

.earthcell-content > .ec-c4 :is( h2, h3 ),
.earthcell-content > .ec-c3 :is( h2, h3 ),
.earthcell-content > .ec-section--rule :is( h2, h3 ),
.earthcell-content > .ec-partners :is( h2, h3 ),
.earthcell-content > .ec-c5 :is( h2, h3 ),
.earthcell-content > .ec-c6 :is( h2, h3 ),
.earthcell-content > .ec-section--narrow :is( h2, h3 ),
.earthcell-content > .ec-c8 :is( h2, h3 ),
.earthcell-content > .ec-c10 :is( h2, h3 ) {
	margin-top: 0;
}

/* And the eyebrow is the first thing in a band, so it carries none either. */

.earthcell-content .ec-eyebrow {
	margin-top: 0;
}

/* The old system's 88px between sections is the band gap now, and bands do not
   stack margins: the colour change is the boundary and it must be exact. */

.earthcell-content > .ec-section,
.earthcell-content > .ec-cta,
.earthcell-content > .ec-band,
.earthcell-content > .ec-partners {
	margin-top: 0;
}

/* --------------------------------------------------------------------------
   BAND 02 REBUILT TO THE DESIGN. My first version was not close to it.

   Piers, with the reference open: "This was the design, very far from what
   you've got." He was right, and the two faults compounded:

   1. I collapsed the map column because the chapter map has never drawn, and
      then spread the trio across three columns to fill the space that left.
   2. The handoff says, in terms, ".ec-trio: single column on this side of the
      grid, gap: 22px". Three across was not a judgement call I was entitled to
      make. It is written down.

   So one decision I did flag turned into a second one I did not, and the band
   stopped resembling the reference at all. A deviation has to stay the size of
   the thing it is about.

   IT IS NOW THE HANDOFF'S LAYOUT: map left at .9fr, words right at 1.1fr, 44px
   between, the trio stacked in the right column with an aqua rule over each
   item.

   THE MAP CONTAINER IS DRAWN AS AN EMPTY PANEL, AND THAT IS STILL NOT RIGHT.
   The widget has never filled data-ec-chapter-map: tested un-hidden at 360px
   with eight seconds to fill, zero children and no canvas, on staging and live
   alike. The reference shows a striped placeholder there, so the design expects
   something. A panel one step darker than the band, at the design's proportion,
   reads as a frame waiting for its picture rather than as a hole. It is not a
   substitute for the map and the widget still owes us one.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c3__inner {
	display: grid;
	grid-template-columns: 0.9fr 1.1fr;
	gap: var(--ec-band-gap);
	align-items: start;
}

.earthcell-content .ec-c3__map {
	display: block;
	aspect-ratio: 1 / 1;
	min-height: 0;
	background: rgba( 5, 8, 11, 0.22 );
	border: 0;
}

/* The trio is ONE column in the right-hand half, which is what the handoff says. */

.earthcell-content .ec-trio {
	display: grid;
	grid-template-columns: minmax( 0, 1fr );
	gap: 22px;
	margin-top: 32px;
}

.earthcell-content .ec-trio__item {
	border-top: 2px solid var(--ec-aqua);
	padding-top: 12px;
}

.earthcell-content .ec-trio__title {
	font-size: 19px;
	line-height: 1.25;
	font-weight: 700;
	margin: 0 0 6px;
	color: var(--ec-paper);
}

.earthcell-content .ec-trio__item p {
	margin: 0;
	font-size: 16px;
	line-height: 1.6;
	color: var(--ec-reversed-body);
}

/*
 * THE CREDIT STAYS THE LAST LINE OF THE WRITTEN COLUMN and must not move into
 * the map. The handoff says so explicitly, and it is the line that names whose
 * imagery the boundary came from.
 */
.earthcell-content .ec-c3__words > .ec-credit {
	margin-top: 24px;
}

@media ( max-width: 900px ) {

	.earthcell-content .ec-c3__inner {
		grid-template-columns: minmax( 0, 1fr );
	}

	.earthcell-content .ec-c3__map {
		aspect-ratio: 4 / 3;
	}
}

/* --------------------------------------------------------------------------
   BANDS 03, 04 AND 05 REBUILT TO THE SPEC, after Piers sent three more
   reference screens. Every value below is the README's.

   THE PATTERN IN ALL THREE MISSES IS THE SAME, and it is worth naming. Each of
   these components was designed for a DARK page, where a label sat over imagery
   in white with a gradient scrim beneath it. On a light band that arrangement
   is wrong twice: the label loses its contrast, and the design puts it UNDER
   the picture in teal instead. Turning the grounds light without re-reading
   each component's own spec is what produced three bands that did not resemble
   the reference at all.
   -------------------------------------------------------------------------- */

/* 03 THE SITES.
   "grid repeat(3, 1fr), gap 28px. Each is a 130px satellite thumbnail, then a
   mono meta line in #005A69, then h3 21px weight 700, then a 16px/1.58
   paragraph. 10px between the parts." */

.earthcell-content .ec-sitecards {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 28px;
}

.earthcell-content .ec-sitecard {
	display: flex;
	flex-direction: column;
	gap: 10px;
}

.earthcell-content .ec-sitecard__diagram {
	aspect-ratio: auto;
	height: 130px;
	overflow: hidden;
}

/* The scrim and the gradient existed to hold white type over the picture. The
   label sits under the picture now, so both would only darken the thumbnail. */
.earthcell-content .ec-sitecard__diagram::after,
.earthcell-content .ec-sitecard__scrim {
	display: none;
}

.earthcell-content .ec-sitecard__scale {
	position: static;
	inset: auto;
	order: 1;
	margin: 0;
	padding: 0;
	font-family: var(--ec-mono);
	font-size: 11px;
	line-height: 1.4;
	font-weight: 600;
	letter-spacing: 0.16em;
	text-transform: uppercase;
	color: var(--ec-teal);
	text-shadow: none;
	background: none;
}

.earthcell-content .ec-sitecard__name {
	order: 2;
	font-size: 21px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: -0.01em;
}

.earthcell-content .ec-sitecard__body {
	order: 3;
	font-size: 16px;
	line-height: 1.58;
}

/* 03 THE PARTNERS.
   "separated from the sites by 48px of space and a single 1px solid
   rgba(62,68,84,.18) rule with 32px above the heading... h3 24px weight 700;
   lede 18px/1.62, max-width 60ch, 24px below; logos grid repeat(3, 1fr),
   gap 24px, each row 64px tall." */

.earthcell-content > .ec-partners > * {
	margin-top: 48px;
	padding-top: 32px;
	border-top: 1px solid rgba( 62, 68, 84, 0.18 );
}

.earthcell-content > .ec-partners h3 {
	margin: 0 0 8px;
	font-size: 24px;
	line-height: 1.2;
	font-weight: 700;
}

.earthcell-content > .ec-partners > * > p:first-of-type {
	margin-bottom: 24px;
	font-size: 18px;
	line-height: 1.62;
	max-width: 60ch;
}

.earthcell-content .ec-partners__row {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 24px;
	margin-top: 0;
}

.earthcell-content .ec-partners__row > * {
	min-height: 64px;
}

/* 04 THE YEAR ROW.
   "border-top 1px solid rgba(62,68,84,.22), 16px above the years, three equal
   columns, mono 13px weight 700." */


.earthcell-content .ec-c5__year {
	font-size: 13px;
	font-weight: 700;
	letter-spacing: 0.16em;
}

/* 05 THE THREE CAPTURE SCALES.
   "Grid repeat(3, 1fr), gap 24px. Each panel: a 170px image, then a mono 12px
   weight 700 title in #CEF17B 14px under it, then a 16px/1.55 sentence 6px
   under that." The panels sit inside the container rather than meeting the band
   edges, which is what stops this reading as a continuation of the hero. */

.earthcell-content .ec-c6__row {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 24px;
}

.earthcell-content .ec-c6__panel {
	display: flex;
	flex-direction: column;
	position: relative;
	min-height: 0;
	padding: 0;
	background: none;
	border: 0;
	overflow: visible;
	gap: 0;
}

/*
 * THE IMAGE BLOCK IS DRAWN AND STILL HAS NO MAP IN IT. Same position as band
 * 02: data-ec-chapter-map has never been filled by the widget, on staging or on
 * live. 170px is the design's height and the darker panel is a frame waiting
 * for its picture. It is not a map, and the widget still owes us three.
 */
.earthcell-content .ec-c6__map {
	display: block;
	position: static;
	inset: auto;
	height: 170px;
	background: rgba( 5, 8, 11, 0.22 );
}

.earthcell-content .ec-c6__caption {
	position: static;
	inset: auto;
	display: block;
	padding: 0;
	background: none;
}

.earthcell-content .ec-c6__label {
	display: block;
	margin: 14px 0 6px;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1;
	font-weight: 700;
	letter-spacing: 0.16em;
	text-transform: uppercase;
	color: var(--ec-lime);
	text-shadow: none;
	background: none;
}

@media ( max-width: 900px ) {

	.earthcell-content .ec-sitecards,
	.earthcell-content .ec-partners__row,
	.earthcell-content .ec-c6__row,
	.earthcell-content .ec-c5__years {
		grid-template-columns: minmax( 0, 1fr );
	}
}

/* --------------------------------------------------------------------------
   THE PAGE HEADING WAS SITTING UNDER THE FIXED SITE HEADER.

   Found by a usability pass, not by me. On /earthcell/what-is-an-earthcell/ at
   390px the header's bottom edge is at 68px and the H1's top at 40px, so a
   visitor reads "EarthCell?" and the words "What is an" are behind the
   navigation. At 1440 the H1 top is 70 against a header bottom of 73, so the
   tops of the letters are clipped. That is the page that owns the phrase.

   The site header is fixed and sits OVER the page rather than above it, so any
   template that opens with type rather than a photograph has to pay for its own
   clearance. calculator-hub.css solves the same problem with --hub-header and
   says so; the article template never did.
   -------------------------------------------------------------------------- */

.earthcell-section.page-article .page-article__head {
	padding-top: 96px;
	scroll-margin-top: 96px;
}

/* --------------------------------------------------------------------------
   WHAT A BAND-BY-BAND COMPARISON AGAINST THE REFERENCE FOUND, AND THE TWO
   MISTAKES BEHIND MOST OF IT.

   MISTAKE ONE: I STYLED CLASS NAMES I HAD GUESSED. Band 05's parts are
   .ec-c6__kind and .ec-c6__statement. I wrote rules for .ec-c6__label and
   .ec-c6__caption, which exist elsewhere in this file, so the rules were valid
   CSS, matched nothing, and the band kept its old treatment while the commit
   message said it was fixed. Read the markup for the class, never infer it from
   a sibling component.

   MISTAKE TWO: MY OWN BAND TEXT RULES OUTRANK THE EYEBROW. A rule reading
   `.earthcell-content > .ec-section--rule :is(h2,h3,p,li)` is two classes and
   an element; `.earthcell-content .ec-eyebrow` is two classes. The first wins,
   so every eyebrow on a light band came out 18px slate instead of 11px teal -
   the single most visible difference from the reference, on four bands.

   The band text rules now exclude the eyebrow explicitly. A rule that colours
   "all the text in this band" has to say what it does not mean.
   -------------------------------------------------------------------------- */

/* THE EYEBROW WINS. Named with the band so it outranks the band's own text
   rules, and set here in full so there is one place to read its specification. */

.earthcell-content > .ec-c4 .ec-eyebrow,
.earthcell-content > .ec-c3 .ec-eyebrow,
.earthcell-content > .ec-section--rule .ec-eyebrow,
.earthcell-content > .ec-partners .ec-eyebrow,
.earthcell-content > .ec-c5 .ec-eyebrow,
.earthcell-content > .ec-c6 .ec-eyebrow,
.earthcell-content > .ec-section--narrow .ec-eyebrow,
.earthcell-content > .ec-c8 .ec-eyebrow,
.earthcell-content > .ec-c10 .ec-eyebrow {
	display: block;
	margin: 0 0 20px;
	max-width: none;
	font-family: var(--ec-mono);
	font-size: 11px;
	line-height: 1;
	font-weight: 600;
	letter-spacing: 0.16em;
	text-transform: uppercase;
	color: var(--ec-teal);
}

.earthcell-content > .ec-c3 .ec-eyebrow,
.earthcell-content > .ec-c6 .ec-eyebrow {
	color: var(--ec-lime);
	margin-bottom: 28px;
}

.earthcell-content > .ec-c8 .ec-eyebrow {
	color: var(--ec-eyebrow-on-lime);
}

/* 02 THE TRIO RULE IS AQUA, NOT LIME. "border-top: 2px solid #00EED2. Aqua on
   teal is a rule, not text, so its contrast is not a reading concern." */

.earthcell-content > .ec-c3 .ec-trio__item {
	border-top: 2px solid #00eed2;
}

/* 03 THE PARTNERS HEADING IS AN h2 IN THE MARKUP, and the spec asks for the
   size of an h3. The markup is not changing, so the size comes to it. */

.earthcell-content > .ec-partners :is( h2, h3 ) {
	font-size: 24px;
	line-height: 1.2;
	font-weight: 700;
	max-width: none;
}

/* 03 AND 06 THE READING MEASURES. Both bands were running their prose the full
   1040px of the container. "62ch" and "60ch" are the spec's, and at 18px they
   land near 640px, which is the measure the rest of the section reads at. */

.earthcell-content > .ec-section--rule > * > p,
.earthcell-content > .ec-partners > * > p {
	max-width: 62ch;
}

/* 06 IS ONE STEP LARGER THAN THE REST OF THE PAGE, "because this is the
   definition the whole page turns on".

   THE STEP IS NOW THE LEDE, AND ONLY THE LEDE. 19px is not on the page's scale
   at all: the Site Pattern handoff sets 20 / 25 / 31 / 39 / 49 / 61 with 18 for
   body, and 19 sat between two of them meaning nothing. So the definition
   sentence takes 20, which is the lede size, and the three paragraphs after it
   take 18 like every other paragraph on the page. The band still opens larger
   than the rest; it just does it with a size the page already uses. */

.earthcell-content > .ec-section--narrow p {
	font-size: 18px;
	line-height: 1.6;
	max-width: 62ch;
}

/* 05 THE REAL CLASS NAMES. .ec-c6__kind is the SATELLITE / DRONE / GROUND
   label; .ec-c6__statement is the sentence under it. */

.earthcell-content > .ec-c6 .ec-c6__kind {
	display: block;
	margin: 14px 0 6px;
	font-family: var(--ec-mono);
	font-size: 12px;
	line-height: 1;
	font-weight: 700;
	letter-spacing: 0.16em;
	text-transform: uppercase;
	color: var(--ec-lime);
	text-shadow: none;
	background: none;
}

.earthcell-content > .ec-c6 .ec-c6__statement {
	margin: 0;
	font-family: inherit;
	font-size: 16px;
	line-height: 1.55;
	font-weight: 400;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-paper);
	text-shadow: none;
}

/* "The three not-yet-flown and no-field-evidence status lines keep their
   current mono treatment at #a9cccf." They are the honest ones on this band. */

/* .ec-c6__note does not exist in the markup and is not named here: that is the
   guess this pass exists to stop making. Only .ec-c6__status is real. */
.earthcell-content > .ec-c6 .ec-c6__status {
	color: var(--ec-reversed-credit);
}

/* 04 THE YEAR ROW IS A PLAIN ROW OF THREE, not a set of chips. The current year
   is ink and the others are the muted tone; nothing is filled. */

/* .ec-years__item does not exist either. Verified against wordpress-content.html. */
.earthcell-content > .ec-c5 .ec-c5__year {
	background: none;
	border: 0;
	padding: 0;
	border-radius: 0;
}

/* --------------------------------------------------------------------------
   THE FIVE EXTRA HAIRLINES THE COMPARISON FOUND.

   "No hairline rules between sections. The change of ground is the separator."
   The band-level rules were removed, but five survived one level down, inside
   .ec-col, .ec-partner and .ec-c10__inner. On a single ground they were
   invisible; on a band they draw a line across a coloured section for no
   reason. The one rule the design keeps is the one dividing the sites from the
   partners inside band 03, which is set above.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c4 .ec-col,
.earthcell-content > .ec-c3 .ec-col,
.earthcell-content > .ec-section--rule .ec-col,
.earthcell-content > .ec-c5 .ec-col,
.earthcell-content > .ec-c6 .ec-col,
.earthcell-content > .ec-section--narrow .ec-col,
.earthcell-content > .ec-c10 .ec-c10__inner {
	border-top: 0;
	border-bottom: 0;
}

/* --------------------------------------------------------------------------
   THE SECTION HAD FOUR LEFT EDGES. MEASURED, NOT GUESSED.

   Every band's content was supposed to sit on one 1040px column, which lands at
   x=200 in a 1440px window. A measurement of the left edge of every direct child
   of every band found four different values:

     x=200   seven bands, via an __inner wrapper          correct
     x=224   the whole of band 04                          24px too far right
     x=24    band 05's eyebrow, band 06's eyebrow and h2   176px too far LEFT
     x=358   band 06's four paragraphs                     158px too far right

   Band 06 carries three of those four edges by itself, and it is the band that
   holds the definition the rest of the page depends on. Its eyebrow and heading
   sit hard against the window's 24px gutter while its prose sits 334px inboard
   of them, so nothing in it lines up with anything.

   This is the horizontal version of the fault Piers described in the vertical
   axis: "the eye of the viewer is constantly getting distracted by imprecision".
   The three causes are below, and each is fixed at its cause rather than by
   nudging the thing that looked wrong.

   HOW TO CHECK IT AGAIN. Read the left edge of every direct child of every band
   and assert they are all equal. Do not read a screenshot for this; 24px of
   difference is invisible by eye and obvious in a number.
   -------------------------------------------------------------------------- */

/* CAUSE 1: BAND 04's INNER WRAPPER STILL HELD THE OLD DARK-PAGE PADDING.

   `padding: clamp(56px, 9vw, 96px) var(--ec-gutter)` dates from when this
   section was a standalone dark page and each chapter paid for its own space.
   The band system now sets that on the section, so this rule was adding a second
   gutter (the 224px edge) and up to 96px of vertical padding INSIDE a band that
   had already padded 72px. Band 04 was the tallest band on the page for no
   reason a reader could see. The wrapper keeps the column and gives the spacing
   back to the band. */

.earthcell-content .ec-c5__inner {
	padding: 0;
	max-width: var(--ec-band);
	margin: 0 auto;
}

/* CAUSE 2: TWO BANDS HAVE NO INNER WRAPPER AT ALL.

   .ec-c6 and .ec-section--narrow put their children straight into the section,
   so those children were laid out against the section's 24px gutter instead of
   the column. The markup is not changing, so the column moves into the section's
   own padding. max() keeps it honest below 1088px, where the column is wider
   than the window and the gutter takes over again. */

/* A PERCENTAGE IN padding-left DOES NOT MEAN WHAT IT LOOKS LIKE HERE, and the
   first attempt at this rule was wrong because of it. Percentage padding resolves
   against the CONTAINING BLOCK's width, not the element's own. These bands are
   full-bleed by `width: 100vw`, but their containing block is the article column,
   so `( 100% - 1040px ) / 2` came out negative, max() took the gutter, and the
   eyebrow stayed hard left at 16px. Measured, not assumed: the deploy went out
   and the edge had not moved.

   A grid has no such trap. The middle track is the column, capped at 1040px; the
   two side tracks share whatever is left and never fall below the gutter. No
   percentage, no viewport unit, and nothing to get wrong when a scrollbar
   appears and 100vw stops matching the visible width. */

.earthcell-content > .ec-c6,
.earthcell-content > .ec-section--narrow {
	display: grid;
	grid-template-columns:
		minmax( var(--ec-gutter), 1fr )
		minmax( 0, var(--ec-band) )
		minmax( var(--ec-gutter), 1fr );
	padding-left: 0;
	padding-right: 0;
}

.earthcell-content > .ec-c6 > *,
.earthcell-content > .ec-section--narrow > * {
	grid-column: 2;
}

/* CAUSE 3: BAND 06's PARAGRAPHS WERE CENTRED BY AUTO MARGINS.

   A 724px paragraph with `margin-inline: auto` inside a 1392px box centres at
   x=358. Centring a reading measure under a left-aligned heading is what put the
   heading and its own prose 334px apart. The measure stays; the centring goes.
   Reading starts where the heading starts. */

.earthcell-content > .ec-section--narrow > * {
	margin-left: 0;
	margin-right: 0;
}

/* --------------------------------------------------------------------------
   BAND 04: THE YEAR ROW IS A RULE WITH THREE COLUMNS, NOT A FILLED CHIP.

   The previous pass wrote this against `.ec-years`, which does not exist. The
   real wrapper is `.ec-c5__years` and the years are <button> elements. That is
   the second time in this section I have styled a class name I inferred from a
   sibling instead of reading the markup, so the rules below name the buttons
   through their wrapper and their data attribute, which is what the markup has.

   THE LIME FILL ALSO OUTRANKED ME, AND WOULD HAVE AGAIN. The rule that fills the
   chosen year reads

     .earthcell-content .ec-c5:not([data-ec-stage]) .ec-c5__year[data-ec-year-button="0"]

   which is five classes and attributes. My replacement was three, so it lost
   silently and 2026 kept a lime block behind it. The selectors below are five as
   well and sit later in the file, which is what actually decides it.

   THE CHOSEN YEAR STILL HAS TO BE VISIBLE, because these are buttons and pressing
   one changes the list underneath. The design says nothing is filled, so the
   chosen year is marked the way a tab is: full ink, and a 2px ink bar sitting on
   the hairline above it. The other two stay muted. Colour alone would not be
   enough to carry that state, so the bar carries it too.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c5 .ec-c5__years {
	display: grid;
	grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	gap: 0;
	margin-top: 16px;
	padding-top: 0;
	border-top: 1px solid rgba( 62, 68, 84, 0.22 );
	background: none;
}

.earthcell-content > .ec-c5 .ec-c5__years .ec-c5__year[data-ec-year-button] {
	position: relative;
	flex: none;
	margin: 0;
	padding: 16px 0 0;
	background: none;
	border: 0;
	border-radius: 0;
	text-align: left;
	font-family: var(--ec-mono);
	font-size: 13px;
	line-height: 1;
	font-weight: 700;
	letter-spacing: 0.16em;
	color: var(--ec-muted-caption);
	cursor: pointer;
}

.earthcell-content > .ec-c5:not( [data-ec-stage] ) .ec-c5__year[data-ec-year-button="0"],
.earthcell-content > .ec-c5[data-ec-stage="0"] .ec-c5__year[data-ec-year-button="0"],
.earthcell-content > .ec-c5[data-ec-stage="1"] .ec-c5__year[data-ec-year-button="1"],
.earthcell-content > .ec-c5[data-ec-stage="2"] .ec-c5__year[data-ec-year-button="2"] {
	background: none;
	border: 0;
	color: var(--ec-ink);
}

/* The bar is drawn ON the hairline, not under it, so the row reads as one rule
   with a marker rather than as two stacked lines. */
.earthcell-content > .ec-c5:not( [data-ec-stage] ) .ec-c5__year[data-ec-year-button="0"]::before,
.earthcell-content > .ec-c5[data-ec-stage="0"] .ec-c5__year[data-ec-year-button="0"]::before,
.earthcell-content > .ec-c5[data-ec-stage="1"] .ec-c5__year[data-ec-year-button="1"]::before,
.earthcell-content > .ec-c5[data-ec-stage="2"] .ec-c5__year[data-ec-year-button="2"]::before {
	content: "";
	position: absolute;
	top: -1px;
	left: 0;
	width: 100%;
	height: 2px;
	background: var(--ec-ink);
}

.earthcell-content > .ec-c5 .ec-c5__year:hover {
	color: var(--ec-ink);
}

/* --------------------------------------------------------------------------
   BAND 05: THE THREE PANELS DID NOT END ON THE SAME LINE.

   Measured: the three sentences run to one, two and two lines, so the status
   line under each one sat at y=365, y=390 and y=391. Three captions at three
   heights across a row of three equal panels is the same imprecision as the
   left edges above, and it is the thing that made this band look unlike the
   reference even though every colour and size in it was right.

   The panel is a flex column, the caption fills what is left of it, and the
   status is pushed to the bottom. The three then end level whatever the
   sentences do, at any width, in any language.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c6 .ec-c6__panel {
	display: flex;
	flex-direction: column;
}

.earthcell-content > .ec-c6 .ec-c6__caption {
	display: flex;
	flex: 1 1 auto;
	flex-direction: column;
}

.earthcell-content > .ec-c6 .ec-c6__status {
	margin-top: auto;
	padding-top: 10px;
}

/* The eyebrow's own bottom margin was set for a band that opens with prose. This
   one opens with a row of pictures, which needs a little more air under the
   label than a paragraph does. */
.earthcell-content > .ec-c6 .ec-eyebrow {
	margin-bottom: 32px;
}

.earthcell-content > .ec-c6 .ec-credit--c6 {
	margin-top: 28px;
}

@media ( max-width: 900px ) {

	.earthcell-content > .ec-c5 .ec-c5__years {
		grid-template-columns: repeat( 3, minmax( 0, 1fr ) );
	}
}

/* The credit under band 05 carries `margin: 0 auto` from the days when it was a
   centred line on a dark page. As a grid child that recentres it inside the
   column, 161px right of the panels it belongs to. It is a caption for the row
   above it, so it starts where the row starts. */

/* The same auto margin is on the row as well, which recentred it 1px off the
   column on a phone. Both bands get the rule, not one child of one of them: a
   band's children start where the band starts, and an inherited centring from
   the dark-page design is never what puts them there. */

.earthcell-content > .ec-c6 > * {
	margin-left: 0;
	margin-right: 0;
}

/* The credit also carried `padding: 0 96px 0 24px` from the dark page, which
   held its text 24px inboard of the panels it captions while its box measured
   as aligned. Padding inside an aligned box is how a 24px error hides from a
   measurement of left edges. */

.earthcell-content > .ec-c6 > .ec-credit--c6 {
	padding-left: 0;
	padding-right: 0;
}

/* --------------------------------------------------------------------------
   THREE MORE INNER WRAPPERS HELD THE SAME DARK-PAGE GUTTER, AND A BOX
   MEASUREMENT COULD NOT SEE THEM.

   .ec-c4__inner, .ec-c3__inner and .ec-c8__inner each carry `max-width: 1040px`
   AND `padding: 0 24px`, so their boxes sit on the column and their contents sit
   24px inside it. Every check that read left edges said those bands were
   correct. Reading where the INK starts - a Range around the first text node,
   which reports glyphs rather than boxes - put bands 01, 02 and 07 at x=216
   against a column at x=192, on desktop and on a phone alike.

   That is the lesson from this pass. Aligned boxes are not aligned type, and
   padding is exactly where a 24px error goes to hide from a measurement.

   The band already owns the gutter. These wrappers own the column and nothing
   else, which is what .ec-c5__inner and .ec-c10__inner already do.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c4__inner,
.earthcell-content .ec-c3__inner,
.earthcell-content .ec-c8__inner {
	padding-left: 0;
	padding-right: 0;
}

/* ==========================================================================
   THE VERTICAL RHYTHM, TAKEN FROM THE DESIGN FILE BY MEASUREMENT.

   Piers sent the same reference twice and I twice reported a fix without
   comparing against it. The third time I opened the handoff's own HTML
   ("EarthCell Section Rhythm.dc.html") in the browser beside the live page and
   read both with the same script. Every number below is the design file's,
   measured, not inferred from a screenshot.

   WHAT THE COMPARISON FOUND. The colours and sizes were right and the SPACING
   was not, by a lot, which is why the two screens look unlike each other while
   every individual value passes:

                       design   mine    out by
     eyebrow to heading   20px   46px   +26px
     heading to lede      16px   26px   +10px
     paragraph gap        12px   26px   +14px
     text to component    32px   36px    +4px

   The cause is one line: .ec-c5__inner is a flex column with `gap: 26px`, so
   every relationship in the band got the same distance whatever it joined. A
   single gap value cannot express a rhythm. The design distinguishes four
   relationships and the band has to as well.

   The four values hold across bands 02, 04, 05 and 07 in the design file, so
   they are the section's rhythm rather than one band's arrangement. They are
   named as tokens here for the same reason Piers asked for one spacing value
   between sections: so the next band does not get a fifth opinion.

   THE 28px CASE IS THE DESIGN'S, NOT A ROUNDING. Where the thing after the
   eyebrow is a row of pictures rather than type, the design opens it to 28px
   (band 05), and band 07 closes to 28px before its buttons. Both are in the
   file and both are kept.
   ========================================================================== */

.earthcell-content {
	--ec-gap-eyebrow: 20px;   /* an eyebrow to what it labels */
	--ec-gap-eyebrow-row: 28px; /* ... when what follows is pictures, not type */
	--ec-gap-heading: 16px;   /* a heading to its first paragraph */
	--ec-gap-para: 12px;      /* one paragraph to the next */
	--ec-gap-block: 32px;     /* type to a component: a row, a grid, buttons */
}

/* A FLAT GAP CANNOT EXPRESS A RHYTHM. The band lays out; the margins space. */

.earthcell-content .ec-c5__inner {
	gap: 0;
}

.earthcell-content > .ec-c5 .ec-eyebrow {
	margin: 0 0 var(--ec-gap-eyebrow);
}

/* The chip is a second label under the eyebrow. The design has no chip here,
   because the design file is abridged and the chip is one of the honesty
   markers that must stay. It reads as part of the label group, so it takes the
   paragraph gap above it and the eyebrow gap below. */
.earthcell-content .ec-c5__head {
	margin: 0;
	gap: 0;
}

.earthcell-content .ec-c5__chip {
	margin: 0 0 var(--ec-gap-eyebrow);
}

.earthcell-content .ec-c5__title {
	margin: 0;
}

.earthcell-content .ec-c5__body {
	margin: var(--ec-gap-heading) 0 0;
	font-weight: 400;
}

.earthcell-content .ec-c5__body + .ec-c5__body {
	margin-top: var(--ec-gap-para);
}

.earthcell-content .ec-c5__grid,
.earthcell-content .ec-c5__strapline {
	margin-top: var(--ec-gap-block);
}

/* --------------------------------------------------------------------------
   THE BAR OVER 2026 IS MINE, AND THE DESIGN DOES NOT HAVE ONE.

   I added a 2px ink bar to mark the chosen year, reasoning that a button needs
   more than colour to show its state. The design file marks it with colour
   alone: the rule is one uniform hairline across all three columns, the chosen
   year is ink and the other two are #6D7583. The dark block over the first
   third is the single most obvious difference between the two screenshots
   Piers put side by side.

   Inventing a component because it seemed necessary is the same fault as
   guessing a class name. If the state needs a stronger cue the design file is
   where that gets decided, not here.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c5:not( [data-ec-stage] ) .ec-c5__year[data-ec-year-button="0"]::before,
.earthcell-content > .ec-c5[data-ec-stage="0"] .ec-c5__year[data-ec-year-button="0"]::before,
.earthcell-content > .ec-c5[data-ec-stage="1"] .ec-c5__year[data-ec-year-button="1"]::before,
.earthcell-content > .ec-c5[data-ec-stage="2"] .ec-c5__year[data-ec-year-button="2"]::before {
	content: none;
}

/* The unchosen years are #6D7583 in the design. I had them at #5D6672, which is
   the caption tone: a different grey, one step darker, so the row read with
   less separation between the chosen year and the others than intended. */
.earthcell-content > .ec-c5 .ec-c5__years .ec-c5__year[data-ec-year-button] {
	color: var(--ec-muted-year);
}

.earthcell-content > .ec-c5 .ec-c5__years {
	margin-top: 0;
	padding-top: 16px;
	border-top: 1px solid rgba( 62, 68, 84, 0.22 );
}

/* --------------------------------------------------------------------------
   BAND 05: TWO VALUES OFF.

   The design puts 6px between the SATELLITE / DRONE / GROUND label and its
   sentence; mine had none, so label and sentence read as one block instead of
   two. And the sentence is #E2EEF0, a step below paper, where mine used paper
   itself at #F4F8F9 - the same weight as a heading would take on this ground.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c6 .ec-eyebrow {
	margin-bottom: var(--ec-gap-eyebrow-row);
}

.earthcell-content > .ec-c6 .ec-c6__statement {
	margin-top: 6px;
	color: var(--ec-reversed-body);
}

/* --------------------------------------------------------------------------
   TWO FAULTS FROM THE COMMIT ABOVE, BOTH CAUGHT BY RE-MEASURING IT.

   ONE: A FLEX GAP ADDS TO MARGINS, IT DOES NOT REPLACE THEM. Making
   .ec-c6__caption a flex column (so the status line could sit at the bottom)
   brought an existing `gap: 11px` into play. The label's 6px bottom margin, the
   gap's 11px and the sentence's 6px top margin summed to 23px where the design
   says 6px. This is the same lesson as the band rhythm one screen earlier: a
   flat gap and a set of margins cannot both own the spacing. The margins own
   it, so the gap goes to zero and the 6px is stated once.

   TWO: MY OWN BLANKET RULE UNSELECTED THE CHOSEN YEAR. The rule setting every
   year to #6D7583 is five classes and attributes and sits later in the file
   than the rule making the chosen one ink, which is also five. 2026 came out
   the same grey as 2027, so the row lost the only mark it has left after the
   bar was removed. The chosen-year rules now name .ec-c5__years too, which is
   six, so they win on weight rather than on position in the file.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c6 .ec-c6__caption {
	gap: 0;
}

.earthcell-content > .ec-c6 .ec-c6__kind {
	margin-bottom: 0;
}

.earthcell-content > .ec-c5:not( [data-ec-stage] ) .ec-c5__years .ec-c5__year[data-ec-year-button="0"],
.earthcell-content > .ec-c5[data-ec-stage="0"] .ec-c5__years .ec-c5__year[data-ec-year-button="0"],
.earthcell-content > .ec-c5[data-ec-stage="1"] .ec-c5__years .ec-c5__year[data-ec-year-button="1"],
.earthcell-content > .ec-c5[data-ec-stage="2"] .ec-c5__years .ec-c5__year[data-ec-year-button="2"] {
	color: var(--ec-ink);
}

/* --------------------------------------------------------------------------
   A DEFENCE AGAINST WORDPRESS WRAPPING A COMMENT IN A PARAGRAPH.

   wpautop treats a comment block between two blocks as a paragraph and wraps it.
   The page carries ninety-one of these. Ninety render at zero height; the one
   inside the Esmeralda card rendered as a 28px paragraph with a 20px margin and
   pushed that card 58px below the two beside it.

   The comments have moved out of the cards, which is the real fix. This rule is
   the guard, so the next comment written inside a card cannot do it again. Every
   real paragraph in a card carries .ec-sitecard__body, so a classless one is
   always the editor's doing and never the author's.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-sitecard > p:not( [class] ) {
	display: none;
}

/* The label now sits under the picture in the markup, where the design puts it,
   so it needs no reordering and no scrim. Both are stated here rather than left
   as rules whose comments describe an arrangement the markup no longer has. */

.earthcell-content .ec-sitecard__scale {
	order: 0;
	position: static;
}

/* THE META LINE WAS PAPER-WHITE ON A PAPER BAND, so moving it out of the picture
   made it visible and unreadable in the same step. Paper was right when it sat
   over a satellite photograph; the design puts it under the picture in teal.
   Named through the card so it outranks the reversed-text rules this component
   still carries from the dark page. */

.earthcell-content .ec-sitecards .ec-sitecard .ec-sitecard__scale {
	color: var(--ec-teal);
	text-shadow: none;
	background: none;
}

/* --------------------------------------------------------------------------
   BAND 04 HAD NO PICTURE, AND THAT IS WHY NOBODY COULD TELL WHAT IT WAS.

   Piers sent a screenshot of this band and said "I don't know what this
   section is. It's odd." He was looking at a heading, two paragraphs, a
   hairline with three years across it, and then four rows of grey text that
   line up with nothing. Read without its picture, the year row looks like three
   column headings and the rows underneath look like a table that has broken.

   THE ROWS ARE A LEGEND FOR A MAP THAT WAS SWITCHED OFF. Both design files put
   this band in two columns: the map on the left, and on the right the year
   buttons, the observation rows, the imagery caption and the credit. The rows
   say what has been recorded over that piece of ground. With the ground gone
   they caption nothing.

   WHY IT WAS SWITCHED OFF, AND WHY THAT REASONING WAS WRONG. An earlier pass
   hid all five chapter maps, with a note saying they had been "tested un-hidden
   at 360px with eight seconds to fill, all five stayed empty, no canvas and no
   children, on staging and on live alike". Measured again on 6 September 2026
   on staging, every one of them mounts: .ec-c3__map carries a 448x448 canvas,
   the three .ec-c6__map panels carry 331x170 canvases, and the page pulls 218
   Esri World Imagery tiles while it scrolls. A later pass restored c3 and c6 to
   display:block and left this one rule behind, so band 04 alone kept a
   display:none whose evidence had already been withdrawn.

   The lesson is the one this file keeps relearning: a hidden element cannot be
   asked whether it would draw. It has no layout box, so the lazy mount has no
   size to mount into and no scroll fraction to fire on. The test proved its own
   premise.

   THE HAIRLINES WERE WRITTEN FOR THE OLD NIGHT BAND. The separators under each
   observation row, and the map's own border, are rgba(230,238,240,.12) and
   .16 — paper at low alpha, correct when this band was #05080B. The band is
   tint now, rgb(231,238,240), so both were drawing a colour within one step of
   the ground they sat on. Four rows with no rule between them is the other half
   of why this read as loose text rather than a record. They take the design's
   own hairline on tint, rgba(62,68,84,.22), the same value as the year row.

   THE FADE IS THE SPEC'S FIGURE. Rows belonging to a year the reader has not
   chosen were at 0.34. EARTHCELL-03-VALUES.json sets outOfYearTreatment to 0.55
   and says why: dimming the colour as well took them to about 11% effective
   alpha. 0.34 on tint is close enough to that to read as damage.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-c5__map {
	display: block;
	border: 1px solid rgba( 62, 68, 84, 0.22 );
}

.earthcell-content .ec-c5__entry {
	border-bottom: 1px solid rgba( 62, 68, 84, 0.22 );
}

.earthcell-content .ec-c5[data-ec-stage="0"] .ec-c5__entry[data-ec-entry-year="1"],
.earthcell-content .ec-c5[data-ec-stage="0"] .ec-c5__entry[data-ec-entry-year="2"],
.earthcell-content .ec-c5[data-ec-stage="1"] .ec-c5__entry[data-ec-entry-year="2"] {
	opacity: 0.55;
}

/* --------------------------------------------------------------------------
   THREE COLOURS MEASURED AGAINST THE GROUND THEY ACTUALLY LAND ON.

   The audit block earlier in this file names the pattern and says it had
   happened three times in one day: "a rule that sets colour by ancestor has to
   stop at the first descendant that sets its own background, and CSS will not
   do that on its own." It happened three more times. Measured on staging on
   6 September 2026, with the translucent layers composited rather than read off
   the declaration:

     .ec-sitecard__scale      teal on the night thumbnail        2.55:1
     .ec-choose__fallback     ink on the night picking map       2.06:1
     figcaption               ink at 0.65 alpha on paper         3.56:1

   The section's own design audit sets the bar at 4.5:1 and says nothing may sit
   under it.

   1. THE SITE CARD LABEL. I set this to teal earlier today, on the grounds that
   the design moves it out of the picture and under it. That is true of the
   cards where it sits under the picture, and it is not true here: on the Kogi
   page it is still a child of .ec-sitecard__diagram, which paints itself night.
   So the teal rule stays for the cards it was written for, and a longer
   selector puts it back to paper wherever it is still inside the picture. This
   is the "stop at the first descendant with its own background" rule, written
   out by hand because there is no other way to say it.

   2. THE PICKING MAP FALLBACK. A rule setting these to paper already exists,
   about a thousand lines above, and it has never worked. It is outranked by
   `.earthcell-content > #enquiry.ec-c8 :is(h2, h3, p, li, label, legend)`,
   which carries an id and therefore beats every class selector in the file
   however long. The override has to name the id too, which is why this one is
   ugly. Adding !important instead would have hidden the reason.

   3. THE FIGURE CAPTIONS on the two Bioestrela pages are ink at 0.65 alpha, the
   exact muted-alpha fault the design audit found and replaced everywhere else
   with a solid #5A6270 at 4.9:1. These two were missed because they come from
   the theme's own caption styling rather than from this file.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-sitecards .ec-sitecard .ec-sitecard__diagram .ec-sitecard__scale {
	color: var(--ec-paper);
}

.earthcell-content > #enquiry.ec-c8 .ec-choose__map :is( p, span, h3, div ),
.earthcell-content > #enquiry.ec-c8 .ec-choose__fallback,
.earthcell-content > #enquiry.ec-c8 .ec-choose__readout,
.earthcell-content > #enquiry.ec-c8 .ec-choose__view-eyebrow {
	color: var(--ec-paper);
}

.earthcell-content figcaption {
	color: #5a6270;
	opacity: 1;
}

/* --------------------------------------------------------------------------
   SECTION 07 WAS DESIGNED ON NIGHT AND IS BUILT ON LIME. NOTHING INSIDE IT WAS
   RE-COLOURED, SO EVERY LINE THAT DIVIDES IT WENT INVISIBLE.

   Piers asked why the contact form and the EarthCell picker look like one
   undivided block. They look like one block because the thing that separates
   them is drawn in a colour nobody can see.

   The design puts the enquiry form in a bordered box: EARTHCELL-02 gives it
   `border: 1px solid rgba(230,238,240,.18)` on a #05080B band. The newer
   section-rhythm file moves band 07 to lime, #CEF17B. The border came across
   unchanged. Paper at 18% alpha on lime is not a line. Measured on staging on
   7 September 2026, on a ground of rgb(206,241,123):

     .ec-panel        the box around the whole form   paper .18   invisible
     .ec-c8__choice   the three scale options         paper .16   invisible
     .ec-c8__chip     the three record chips          paper .20   invisible
     .ec-button--lime the submit button               #CEF17B     lime on lime

   The submit button is the clearest one: it is the same colour as the band it
   sits on, so only its near-black label shows and it reads as a line of bold
   text somebody left under the form. It is a button, and it has always been a
   button.

   THE BUTTON INVERTS RATHER THAN DARKENING ITS EDGE. On a paper band the
   primary button is lime with ink text. On a lime band that arrangement has
   nowhere to go, so the fill and the text swap: night ground, lime label. Same
   two colours, same button, the one arrangement that works on this ground.

   WHAT THIS DOES NOT DO. It does not move the picker. The picker appears in
   none of the nine sections of the design reference, and the design's section
   07 is a heading, two paragraphs and this one bordered form. Whether the
   picker belongs here, or above in a numbered section of its own, is a layout
   decision and it is Piers's. This change only makes the division the design
   already specifies visible on the ground the band actually has.
   -------------------------------------------------------------------------- */

/* --------------------------------------------------------------------------
   SECTION 07 REBUILT TO THE TONAL BANDS DESIGN OF 7 SEPTEMBER 2026.

   Piers asked yesterday why the contact form and the EarthCell picker looked
   like one undivided block, and I answered that the design never put them
   together and offered to split them into two sections. That was wrong, and the
   new design file settles it: "EarthCell Page - Tonal Bands.dc.html" keeps both
   in section 07 and separates them a different way. It does it with PAPER
   CARDS on the lime band rather than with borders or with distance.

   WHAT THE DESIGN ACTUALLY SPECIFIES, and what was built instead:

     the readout beside the map   #F4F8F9 paper card    was near-black
     the enquiry form             #F4F8F9 paper card    was no ground at all
     form and record chips        two columns, side by side   was stacked
     the record chips             a vertical stack            was a wide row
     both rows                    auto-fit minmax(320px,1fr)  was 680/340
     the submit button            #005A69 teal, paper text    was lime on lime

   So the reason it read as one block is now clear and it was never the
   arrangement. Two of the three cards had no ground, so there was nothing to
   see the seams of. Give them the paper the design asks for and the section
   reads as four distinct things on a lime field.

   MY OWN FIX FROM YESTERDAY IS REMOVED BY THIS BLOCK. I made the submit button
   near-black with a lime label, reasoning that the lime band left the primary
   button nowhere to go. The design says #005A69, which is the same teal as the
   eyebrow above it and needed no invention. A border I could not check against
   a design file was a guess wearing the clothes of a fix.

   THE PANEL FLIP IS THE DANGEROUS PART OF THIS CHANGE. The readout card was
   built for a near-black ground, so everything in it was written light: the
   text, the "Tell us what you have in mind" link in lime, and the hairlines
   between its rows in paper at 14% alpha. Turning the card paper without
   turning those would have hidden the lot. Each one is named below.
   -------------------------------------------------------------------------- */

/* ROW A: the map and its readout, equal columns, stretched to the same height. */

.earthcell-content > #enquiry.ec-c8 .ec-choose {
	grid-template-columns: repeat( auto-fit, minmax( 320px, 1fr ) );
	gap: 20px;
	align-items: stretch;
	margin-bottom: 40px;
}

.earthcell-content > #enquiry.ec-c8 .ec-choose__panel {
	background: #f4f8f9;
	padding: 24px;
	border: 0;
}

/* Everything the dark card wrote light. The map keeps its own night ground, so
   this is named through the panel and never through .ec-choose. */

.earthcell-content > #enquiry.ec-c8 .ec-choose__panel :is( p, span, div, h3, strong, dt, dd ) {
	color: var(--ec-ink);
}

.earthcell-content > #enquiry.ec-c8 .ec-choose__panel .ec-choose__tell {
	color: var(--ec-teal);
	border-color: rgba( 0, 90, 105, 0.45 );
}

/* Paper at 14% alpha on a paper card is not a line. The design's divider. */

.earthcell-content > #enquiry.ec-c8 .ec-choose__panel :is( .ec-choose__summary, .ec-choose__row, .ec-choose__action ) {
	border-top-color: rgba( 62, 68, 84, 0.18 );
}

.earthcell-content > #enquiry.ec-c8 .ec-choose__panel .ec-choose__go {
	background: var(--ec-teal);
	color: var(--ec-paper);
}

/* ROW B: the form and the record chips, side by side.

   The measure cap comes off this row on purpose. It is a two-column layout and
   not a paragraph, and --ec-measure held it to 680 while the row above it ran
   to 1040, which is what made the picker look bolted on. */

.earthcell-content > #enquiry.ec-c8 .ec-c8__grid {
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 320px, 1fr ) );
	gap: 28px;
	align-items: start;
	max-width: none;
}

.earthcell-content > #enquiry.ec-c8 .ec-panel {
	background: #f4f8f9;
	padding: 28px;
	border: 0;
}

.earthcell-content > #enquiry.ec-c8 .ec-c8__records {
	padding-top: 4px;
}

/* A vertical stack in the right-hand column, not a row across the page. */

.earthcell-content > #enquiry.ec-c8 .ec-c8__chips {
	flex-direction: column;
	gap: 10px;
}

.earthcell-content > #enquiry.ec-c8 .ec-c8__chip {
	width: 100%;
	padding: 14px 16px;
	background: rgba( 244, 248, 249, 0.55 );
	border: 1px solid rgba( 62, 68, 84, 0.25 );
}

/* THE SCALE OPTIONS STACK: the block diagram above its label, not beside it.
   Side by side, the label had 74px and "Part of an EarthCell" broke over three
   lines in a box built for two. */

/*
 * THREE TILES, AND A BARE `1fr` PUSHED THE THIRD ONE OFF A 320px SCREEN.
 *
 * This rule was written as `repeat( 3, 1fr )` and it overrode the earlier rule
 * that had it right. A bare `1fr` track carries an automatic minimum of
 * min-content, so it cannot shrink below the widest thing in it. "A larger
 * landscape" over a 4x4 diagram has a min-content of about 97px, so at 320px
 * the three tiles needed 315 inside a 232px column and the third one ran 30px
 * past the edge of the screen. Measured on staging on 7 September 2026.
 *
 * The handoff's own value is `repeat( auto-fit, minmax( 140px, 1fr ) )`, which
 * folds to one column when three no longer fit rather than overflowing, and it
 * is used here instead of patching the 3 to a minmax. The gap follows the
 * handoff too: one gutter, like every other column gap on the page.
 */
.earthcell-content > #enquiry.ec-c8 .ec-c8__scales {
	grid-template-columns: repeat( auto-fit, minmax( 140px, 1fr ) );
	gap: 12px;
}

.earthcell-content > #enquiry.ec-c8 .ec-c8__choice {
	flex-direction: column;
	align-items: stretch;
	gap: 10px;
	padding: 12px;
	border-color: rgba( 62, 68, 84, 0.3 );
}

.earthcell-content > #enquiry.ec-c8 .ec-c8__choice-title {
	font-size: 14px;
	line-height: 1.4;
	font-weight: 600;
}

.earthcell-content > #enquiry.ec-c8 .ec-button--lime {
	background: var(--ec-teal);
	color: var(--ec-paper);
}


/* ==========================================================================
   THE BAND SEQUENCE, TO THE KOGI SITE PATTERN HANDOFF OF 7 SEPTEMBER 2026.

   The third handoff in three days, and it supersedes the two before it. The
   section-rhythm direction (flat tonal bands, no photography) is marked
   superseded in its own README and kept only as a record of what was rejected.

   ITS CENTRAL FINDING: on this site, photography carries the section
   transitions. /regeneration/bioestrela/ uses the teal ground five times
   without it ever reading as arbitrary, because full-bleed photograph rows sit
   between the colour bands. The colour is not what does the separating.

   The rule it sets: EVERY BOUNDARY IS MARKED EXACTLY ONCE, by a change of
   ground, by a photograph, or by a hairline. Never two of those, and never
   none. Measured on staging before this change, against that rule:

     band  ground now      wanted        boundary above
     04    paper           paper         ok
     03    teal            teal          ok
     05    paper           paper         ok
     06    tint #E7EEF0    paper         WRONG COLOUR, and then no boundary
     07    teal            teal          ok
     08    paper           LIME          UNMARKED: paper meets paper
     09    lime            NIGHT         UNMARKED: lime would meet lime
     10    tint #E7EEF0    paper         wrong colour

   THE PALE TINT IS DELETED AND MUST NOT COME BACK. #E7EEF0 existed for one
   reason: to stop two paper bands touching. That job now belongs to the
   photographic bands and to a single hairline, so the colour has no purpose.
   Two near-identical greys were never a rhythm; they were an apology for not
   having one.

   THE ONE HAIRLINE. Bands 05 and 06 are the only paper-to-paper join left on
   the page, so they get the only hairline: 1px of slate at 16%, full bleed,
   marking the band edge rather than the content. It means "section boundary"
   and it is used nowhere else. Component rules inside a panel are a different
   thing and stay at panel width.

   THE 104px MARGIN ON BAND 09 IS REMOVED. Bands butt directly and space each
   other with padding; a margin between two coloured grounds paints the page
   ground through the join, which is a third thing at the boundary and reads as
   a gap rather than an edge. Every other band already sits at margin zero.

   WHAT THIS BLOCK DOES NOT DO, and what is still owed:

   1. The two photograph bands. Band 03 (the Duanama boulder) and band 10 (the
      closing "Let's talk" ground) are what the handoff says does most of the
      separating. The Duanama image is already on the page inside .ec-c4__photo
      at 405x304 and could be promoted to a full-bleed band in CSS. The closing
      photograph is not on the page at all and needs the markup.
   2. The eyebrows. The handoff rejects the numbered uppercase monospace
      labels ("06 · WHAT IT MEANS") outright: the site does not number its
      sections and has no monospace anywhere. Those strings are copy in the page
      body, not CSS, so they are not mine to change here.
   3. Bands 09 and 10 are one section in this markup. The handoff makes them
      two, with their own headings. That needs the body, not the stylesheet.
   ========================================================================== */

/* The tint is deleted. Both bands that carried it take paper. */

.earthcell-content > .ec-c5,
.earthcell-content > .ec-c10 {
	background: var(--ec-paper);
}

/* The one paper-to-paper join on the page, and the only hairline. */

.earthcell-content > .ec-c5 {
	border-top: 1px solid rgba( 62, 68, 84, 0.16 );
}

/* Band 08 is lime. Slate ink, because paper on lime is 1.2:1. */

.earthcell-content > .ec-section--narrow {
	background: var(--ec-lime);
}

.earthcell-content > .ec-section--narrow :is( h2, h3, p, li, span, strong ) {
	color: var(--ec-ink);
}

.earthcell-content > .ec-section--narrow :is( .ec-eyebrow, .ec-marker ) {
	color: #28513f;
}

/* Band 09 is night, which is what gives the lime band above it an edge.

   The selector carries the id because .earthcell-content > #enquiry.ec-c8
   already sets lime and would otherwise win on specificity. */

.earthcell-content > #enquiry.ec-c8 {
	margin-top: 0;
	background: #05080b;
}

.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner > :is( h2, p, .ec-c8__title, .ec-c8__body ) {
	color: var(--ec-paper);
}

.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner > :is( .ec-eyebrow, .ec-marker ) {
	color: var(--ec-lime);
}

/* THE CHOOSER CHIPS COME OFF, which the handoff asks for and gives its reason:
   with the map rendering they duplicate a control that already exists, and they
   made the band read as four loose panels stacked rather than two. The enquiry
   card then has the row to itself, on the 6-column line. */

.earthcell-content > #enquiry.ec-c8 .ec-c8__records {
	display: none;
}

.earthcell-content > #enquiry.ec-c8 .ec-c8__grid {
	grid-template-columns: minmax( 0, 588px );
}

/* --------------------------------------------------------------------------
   TWO NUMBERS THE SITE PATTERN HANDOFF SETTLES.

   THE UNCHOSEN YEARS. I asked Piers yesterday which document won: the
   section-rhythm design file put the unchosen years at #6D7583, and the design
   audit said nothing on the page may sit under 4.5:1. Measured on the tint they
   were 3.95:1, and on paper after the tint came off, 4.34:1. Still short.

   This handoff answers it without anybody having to choose. Its colour table
   gives one value for "muted caption, credit and inactive-year ink on paper",
   #5d6672, and measures it at 5.4:1. The same tone already carries the captions
   and the prototype note, so the unchosen years stop being a special case.

   THE SITES-TO-PARTNERS JOIN. It is a division inside band 05, not a boundary
   between bands, and the handoff is explicit that it is marked by space and a
   sub-heading rather than by a rule: "The divider that used to sit above
   Project partners inside band 5 is removed: that division is now 96px of space
   and its own 31px sub-heading." It was at 48.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c5 .ec-c5__years .ec-c5__year[data-ec-year-button] {
	color: #5d6672;
}

/* 96, not 48. The band already carried a 48px margin here, so writing 48 again
   changed nothing: measured before and after at the same 48px. The handoff asks
   for twice that, because this is the join doing the work of a divider that was
   taken away. */

.earthcell-content > .ec-partners {
	margin-top: 96px;
}

/* --------------------------------------------------------------------------
   THE EYEBROWS BECOME THE SITE'S OWN h4 PATTERN.

   The Site Pattern handoff rejects the numbered uppercase monospace labels
   outright, and its reason is not a matter of taste: "That was invented here;
   the site does not number its sections and has no monospace anywhere." It is
   the same fault as the bar over 2026 and the pale tint — a component made up
   on this page to solve a problem the site had already solved.

   So they become what /inspiration/ and /regeneration/bioestrela/ use: a plain
   sentence-case phrase at 16px weight 700, in Epilogue, teal on paper, lime on
   dark, #28513f on lime. No number, no uppercase, no letter-spacing, no mono.

   The strings changed with them, in the page body. Four were already the right
   words wearing the wrong clothes — "02 · HOW EARTHCELL WORKS HERE" is "How
   EarthCell works here". Three were not, and the handoff gives each one:

     band 02  WHY THIS MATTERS NOW    ->  The Heart of the World
     band 07  THREE CAPTURE SCALES    ->  How the land is recorded
     band 09  THE ASK                 ->  Let's talk

   BAND 02 ALSO STOPPED SAYING THE SAME SENTENCE TWICE. Its heading was "The
   Heart of the World" and its opening line was "The Kogi know the Sierra Nevada
   de Santa Marta as the Heart of the World." The handoff makes the old heading
   the eyebrow and the opening line the heading, which is the same words in a
   better order and one fewer repetition. No word on the page is new.

   BAND 07 HAD AN EYEBROW AND NO HEADING, so its eyebrow text was carrying the
   heading's job at 11px. "Three capture scales" is now the h2 it always was.

   MONOSPACE SURVIVES IN EXACTLY TWO PLACES, both in the sponsor record panel:
   the cell code and its sample status line. Those are machine values and
   monospace says so. Everywhere else it goes.
   -------------------------------------------------------------------------- */

/* THE CLASS IS NAMED TWICE ON PURPOSE. An earlier rule scopes the eyebrow
   through each band — `.earthcell-content > .ec-c4 .ec-eyebrow` and seven more
   like it — which is three classes to this rule's two, so specificity beat
   source order and the first version of this block changed nothing at all: the
   copy came through in sentence case and rendered at 11px uppercase monospace
   anyway. Repeating the class matches that weight without copying an
   eight-selector list that would then have to be kept in step with it. */

/* THE CLASS IS NAMED THREE TIMES NOW, AND HERE IS THE THIRD REASON.

   Two classes were not enough either. Band 08 carries
   `.earthcell-content > .ec-section--narrow > p:first-of-type { font-size: 19px }`,
   which is three classes and an element to this rule's three classes, so it
   won, and the eyebrow "What it means" rendered at 19px while the other six
   rendered at 16. The rule was written when the section's first paragraph was
   its lede. Adding an eyebrow above the heading made the eyebrow the first
   paragraph, and the lede stopped getting the size that was meant for it.

   That rule is corrected below to name the lede rather than whatever happens to
   come first. The third class here is belt as well as braces: every eyebrow on
   the page now outranks any band-level rule that selects a bare `p`, so the
   next band that gets its own paragraph size cannot silently take one of them
   with it.

   THE TYPEFACE IS NAMED RATHER THAN INHERITED. `inherit` took the body font,
   which on this site is Mona Sans, so all seven eyebrows rendered in it while
   every heading beside them was Epilogue. The handoff asks for Epilogue and the
   headings already use exactly this stack. */

.earthcell-content .ec-eyebrow.ec-eyebrow.ec-eyebrow {
	margin: 0 0 20px;
	font-family: Epilogue, sans-serif;
	font-size: 16px;
	line-height: 1;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-teal);
}

/* The three grounds an eyebrow can land on. */

.earthcell-content > .ec-c3 .ec-eyebrow.ec-eyebrow,
.earthcell-content > .ec-c6 .ec-eyebrow.ec-eyebrow,
.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner > .ec-eyebrow.ec-eyebrow {
	color: var(--ec-lime);
}

.earthcell-content > .ec-section--narrow .ec-eyebrow.ec-eyebrow {
	color: #28513f;
}

/* Band 07's new heading takes the same h2 treatment as every other band's. */

.earthcell-content > .ec-c6 .ec-c6__title {
	margin: 0 0 24px;
	max-width: 20ch;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	color: var(--ec-paper);
	text-wrap: pretty;
}

/* ==========================================================================
   THE TWO PHOTOGRAPH BANDS, AND THE SPLIT OF THE OLD .ec-c8 INTO TWO.

   The Kogi Site Pattern handoff of 7 September 2026 owed three things after the
   band sequence landed. This block pays all three.

   1. BAND 03, THE DUANAMA PHOTOGRAPH, FULL BLEED. It was a 405px figure inside
      band 02.
   2. BAND 10, THE CLOSING PHOTOGRAPH, under the enquiry panel. It was not on
      the page at all.
   3. BANDS 09 AND 10 WERE ONE SECTION. They are now .ec-c9 and .ec-c8.

   WHY THE PHOTOGRAPHS ARE STRUCTURE AND NOT DECORATION, in the handoff's own
   words: "on this site, photography carries the section transitions."
   /regeneration/bioestrela/ uses teal five times without it reading as
   arbitrary, because full-bleed photograph rows sit between the colour bands.
   The rule that follows is that every boundary is marked exactly once. Take
   either photograph away and four boundaries lose their only marker:

     02 -> 03  photograph          the Duanama band
     03 -> 04  photograph to teal
     09 -> 10  photograph          the closing band
     10 -> 11  photograph to paper

   THE CONTAINER IS THE PAGE'S OWN 1040, NOT THE HANDOFF'S 1200. The handoff
   specifies a 1200px container and a 78px column, and the whole page would have
   to move to it at once: every band, every track, every alignment line. That is
   a separate change. Until it happens, a caption or a track written to 1200
   would sit 80px outside every other line on the page, which is worse than
   being on a narrower grid consistently. So the ratios below are the handoff's
   exactly (690/1176 and 486/1176) and the container is --ec-band.

   THE TRACK WIDTH GOES IN THE FLEX BASIS, NEVER IN flex-grow OFF A ZERO BASIS.
   The handoff measured that fault and gives the reason: with flex-basis 0 a
   padded track's padding lands outside its grown share, so any panel with
   padding ends up wider than its proportion. Percentage bases include padding
   under border-box, so the two bases sum to the space and the ratio holds
   whatever the padding is.
   ========================================================================== */

/* --------------------------------------------------------------------------
   BAND 03 AND THE PHOTOGRAPH BAND PATTERN.

   THE FULL-BLEED SUM IS THE FILE'S OWN, used by .ec-band and by every other
   band here, and it is used again rather than a cleverer one. See the long note
   beside the band list: an invalid calc does not fall back, it reverts the
   property to 0, and the band then keeps its container's left edge.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-photoband {
	position: relative;
	display: flex;
	flex-direction: column;
	justify-content: flex-end;
	width: 100vw;
	margin: 0 0 0 calc( 50% - 50vw );
	padding: 0;
	min-height: clamp( 280px, 38vw, 480px );
	overflow: hidden;
	border: 0;

	/* The ground shows only if the photograph fails to load. It is still worth
	   having: a band that is meant to be a picture should not be a hole. */
	background: #16242a;
}

.earthcell-content > .ec-photoband > .ec-photoband__img {
	position: absolute;
	inset: 0;
	display: block;
	width: 100%;
	height: 100%;

	/* The band list gives every direct child of a band max-width: var(--ec-band)
	   and auto margins. That is right for words and wrong for a ground, so both
	   are undone here. */
	max-width: none;
	margin: 0;
	object-fit: cover;
}

/*
 * THE CAPTION CARRIES ITS OWN SCRIM, RATHER THAN THE BAND CARRYING ONE.
 *
 * A scrim over the whole picture would darken the middle of it, which is the
 * fault the handoff records against the hero's first top-to-bottom gradient.
 * This one is a bottom-up fade inside the last 68px, so the boulder and the
 * people at it are clear and the caption is still legible at 10.2:1.
 *
 * THE PADDING PUTS THE CAPTION ON THE CONTAINER LINE. max() rather than a
 * media query: the gutter wins on a phone, the centring sum wins on a desktop,
 * and the left edge never moves at any width in between.
 *
 * THE SCRIM IS STRONGER THAN THE HANDOFF'S, AND HERE IS THE MEASUREMENT. The
 * handoff gives rgba(5,8,11,.8) fading to nothing, and records the caption at
 * 10.2:1. That is not what this photograph produces. The caption line sits
 * about 44% of the way up the scrim, where a straight .8-to-0 fade has only
 * reached .44, and the Duanama boulder is a bright pale grey exactly there.
 * Measured over the real image at four viewport widths, composited properly:
 * 1.75, 2.06, 2.23 and 2.25 to one. Against a floor of 4.5.
 *
 * So the fade keeps its shape and gains a plateau: .92 at the bottom edge,
 * still .78 at 55% where the words are, then away to nothing at the top. Same
 * four widths, re-measured: 6.24, 6.61, 6.82 and 6.82 to one. The scrim covers
 * the bottom 88px of a 480px band, so the boulder and the people at it are
 * still clear, which is the whole reason the picture is a band and not a
 * background.
 */
.earthcell-content > .ec-photoband > .ec-photoband__caption {
	position: relative;
	max-width: none;
	margin: 0;
	padding: 48px max( var(--ec-gutter), calc( ( 100% - var(--ec-band) ) / 2 ) ) 20px;
	font-family: inherit;
	font-size: 13px;
	line-height: 1.5;
	font-weight: 400;
	letter-spacing: normal;
	text-transform: none;
	color: #a9bcc0;
	background: linear-gradient( 0deg, rgba( 5, 8, 11, 0.92 ) 0%, rgba( 5, 8, 11, 0.78 ) 55%, rgba( 5, 8, 11, 0 ) 100% );
}

/* --------------------------------------------------------------------------
   BAND 02 LOSES ITS PHOTOGRAPH AND TAKES THE HANDOFF'S HEADING / PROSE SPLIT.

   It was a two-column grid at 1.15fr / 0.85fr with the words in one column and
   the photograph in the other. With the photograph promoted to a band, that
   grid had one child and half the band was empty.

   The handoff's grid section gives two row shapes. This is the first of them:
   5 columns of heading, 7 of prose, eyebrow and heading left, body copy right.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c4 .ec-c4__inner {
	display: flex;
	flex-wrap: wrap;
	gap: var(--ec-gutter);
	align-items: start;
	grid-template-columns: none;
}

.earthcell-content > .ec-c4 .ec-c4__words {
	flex: 1 1 calc( ( 100% - 24px ) * 486 / 1176 );
	min-width: min( 300px, 100% );
}

.earthcell-content > .ec-c4 .ec-c4__prose {
	flex: 1 1 calc( ( 100% - 24px ) * 690 / 1176 );
	min-width: min( 340px, 100% );
}

/* The measure goes on the type, never on the track. */

.earthcell-content > .ec-c4 .ec-c4__title {
	max-width: none;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	text-wrap: pretty;
}

.earthcell-content > .ec-c4 .ec-c4__prose .ec-c4__body {
	max-width: none;
	font-size: 18px;
	line-height: 1.6;
}

/* The band opens on the lede size, which is one step above body. */

.earthcell-content > .ec-c4 .ec-c4__prose .ec-c4__body:first-child {
	font-size: 20px;
}

.earthcell-content > .ec-c4 .ec-c4__prose > * + .ec-c4__body {
	margin-top: 16px;
}

/* --------------------------------------------------------------------------
   BAND 09, CHOOSE AN EARTHCELL. NIGHT.

   This is the half of the old .ec-c8 that holds the map of cells. It takes the
   night ground the old section had, which is what gives the lime band above it
   an edge. The enquiry keeps the id, because #enquiry is the anchor a reader
   follows to write to us, not to look at a grid.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c9 {
	width: 100vw;
	margin: 0 0 0 calc( 50% - 50vw );
	padding: var(--ec-band-pad) var(--ec-gutter);
	border: 0;
	background: #05080b;
	color: var(--ec-paper);
}

.earthcell-content > .ec-c9 > * {
	max-width: var(--ec-band);
	margin-left: auto;
	margin-right: auto;
	padding-top: 0;
	padding-bottom: 0;
}

.earthcell-content > .ec-c9 :is( h2, h3, p, li ) {
	color: var(--ec-paper);
}

.earthcell-content > .ec-c9 .ec-eyebrow.ec-eyebrow {
	color: var(--ec-lime);
}

.earthcell-content > .ec-c9 .ec-c9__title {
	margin: 0 0 48px;
	max-width: 486px;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	color: var(--ec-paper);
	text-wrap: pretty;
}

/* --------------------------------------------------------------------------
   BAND 10, THE ASK, ON THE CLOSING PHOTOGRAPH.

   THE SCRIM RUNS LEFT TO RIGHT, not top to bottom, which is the hero's
   treatment and the handoff's reason for it: a top-down scrim darkens the
   middle of the picture. Here it also has to hold the words beside the enquiry
   panel legible on a photograph that is bright at the top.

   THE GROUND UNDER THE PHOTOGRAPH IS #1c2c26, the handoff's own value for this
   band, so the band still reads as a band if the image never arrives.
   -------------------------------------------------------------------------- */

.earthcell-content > #enquiry.ec-c8 {
	position: relative;
	overflow: hidden;
	background: #1c2c26;
}

.earthcell-content > #enquiry.ec-c8 > .ec-c8__photo {
	position: absolute;
	inset: 0;
	z-index: 0;
	display: block;
	width: 100%;
	height: 100%;
	max-width: none;
	margin: 0;
	object-fit: cover;
}

.earthcell-content > #enquiry.ec-c8::before {
	content: "";
	position: absolute;
	inset: 0;
	z-index: 1;
	background: linear-gradient( 90deg, rgba( 5, 20, 15, 0.94 ) 0%, rgba( 5, 20, 15, 0.86 ) 45%, rgba( 5, 20, 15, 0.4 ) 80%, rgba( 5, 20, 15, 0.2 ) 100% );
}

.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner {
	position: relative;
	z-index: 2;
}

/* The lede sits at the handoff's 20px, and on the reversed tone for night. */

.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner > .ec-c8__body {
	max-width: 588px;
	font-size: 20px;
	line-height: 1.6;
	color: #dbe6e8;
}

.earthcell-content > #enquiry.ec-c8 > .ec-c8__inner > .ec-c8__title {
	max-width: 486px;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	text-wrap: pretty;
}

/* ONE PANEL INSET ON THE PAGE, and this is it. The handoff records three
   different insets (24, 32, 40) as the single largest source of drift, because
   they put panel text on three different lines while every other paragraph
   started on the container line. This panel was at 28. */

.earthcell-content > #enquiry.ec-c8 .ec-panel {
	padding: clamp( 20px, 3vw, 32px );
}

/* --------------------------------------------------------------------------
   BAND 04 PUTS ITS WORDS ON THE CONTAINER LINE, WHERE EVERY OTHER BAND'S ARE.

   THE FAULT, MEASURED. The handoff's alignment table has one line that carries
   fifty elements: offset 0, the container's left edge, "every eyebrow, heading
   and paragraph that opens a block". Measured on staging at 1920px, seven of
   the page's eight eyebrows sat at x=432. Band 04's sat at x=925, because the
   map held the left half of the row and the words held the right.

   THE RATIO WAS WRONG AS WELL. The row was a 0.9fr / 1.1fr grid with a 44px
   gap, which rendered 448 and 548: a ratio of 1.223 where the handoff's
   feature / aside split is 1.420. It is now the handoff's own basis sum, and
   the gap is one gutter like every other column gap on the page.

   THE ORDER IS SWAPPED IN CSS, NOT IN THE PAGE BODY, and that is safe here for
   one specific reason: .ec-c3__map is empty and carries aria-hidden="true", so
   it is not in the reading order at all. Swapping two flex items would
   normally split what a screen reader hears from what a sighted reader sees.
   Here there is nothing to split. The data-ec-chapter-map attribute the widget
   mounts on is untouched.

   THE MAP STAYS, AND THAT IS A DECISION, NOT AN OMISSION. The handoff puts a
   facts panel in this aside instead: five label-over-value pairs for Location,
   Related sites, Project partners, Certification and Imagery. Asked on
   7 September 2026, Piers kept the map: "Three simple principles currently has
   a nice image with cells over it. Let's keep that there."

   So this is the one place the page departs from the handoff on purpose. The
   map shows the actual cells over the actual land, which is the thing the
   three principles are about, and a list of five facts is not. Do not propose
   the facts panel here again without asking him first.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-c3 .ec-c3__inner {
	display: flex;
	flex-wrap: wrap;
	gap: var(--ec-gutter);
	align-items: start;
	grid-template-columns: none;
}

.earthcell-content > .ec-c3 .ec-c3__words {
	order: 1;
	flex: 1 1 calc( ( 100% - 24px ) * 690 / 1176 );
	min-width: min( 340px, 100% );
}

.earthcell-content > .ec-c3 .ec-c3__map {
	order: 2;
	flex: 1 1 calc( ( 100% - 24px ) * 486 / 1176 );
	min-width: min( 300px, 100% );
}

/*
 * THE STRAY EMPTY PARAGRAPH GOES BACK TO ZERO WIDTH. WordPress writes a bare
 * <p></p> wherever the stored body has a blank line, and .ec-c3__inner has one
 * as its first child. In a grid it took a cell and nobody noticed. In a flex
 * row it takes a basis and pushes the tracks off their ratio.
 */
.earthcell-content > .ec-c3 .ec-c3__inner > p:empty {
	display: none;
}

/* ==========================================================================
   MONOSPACE COMES OFF EVERYTHING EXCEPT THE CELL CODES.

   The Site Pattern handoff allows monospace in exactly two places, both inside
   the sponsor record panel: the cell code (EC-KOGI-000027) and its status line
   (SAMPLE · PART-FUNDED). Its reason is not taste. Codes and sample markers are
   machine values, and monospace says so. Nothing else on the page is a machine
   value, so nothing else earns it.

   Measured on staging before this block: thirty-seven elements were monospace
   or uppercase or both. Six were the codes and sample markers. The rest were
   captions, credits, statuses, field labels, kind labels and a chip, all
   wearing a typeface that told the reader they were data.

   THE STRINGS CHANGED WITH THEM, IN THE PAGE BODY, and every replacement is the
   reference file's own wording read off the rendered prototype. "IMAGERY ©
   ESRI, MAXAR · LA ESMERALDA PROJECT BOUNDARY" is "Imagery © Esri, Maxar · La
   Esmeralda project boundary". Same words, and no word invented. A
   text-transform cannot undo capitals that are in the copy, which is why this
   could not be a stylesheet change on its own.

   THE THREE CAPTURE-SCALE LABELS BECAME REAL HEADINGS. Satellite, Drone and
   Ground were paragraphs styled to look like headings, inside a band whose h2
   is "Three capture scales". They are h3 now, at the handoff's 25px, so the
   page's outline says what its type already said. Nothing in the widget or the
   tests refers to them, checked before changing the tag.

   THE ONE LABEL THAT CHANGED ROLE. "DISCUSS A CORPORATE CONTRIBUTION" was a
   10px marker over the enquiry form. The handoff makes it the panel's h3 at
   25px, which is what it is: the panel has a heading, and a heading is not a
   caption in capitals.

   THE BREADCRUMB IS LEFT ALONE. It is IBM Plex Mono at 12px and it is the
   site's own breadcrumb, used on pages this handoff says nothing about.
   Changing it here would change them.
   ========================================================================== */

/* Captions, credits and statuses. 13px Epilogue, coloured by their ground. */

.earthcell-content .ec-credit,
.earthcell-content .ec-c4__note,
.earthcell-content .ec-sitecards__note,
.earthcell-content .ec-c6__status,
.earthcell-content .ec-c6__caption,
.earthcell-content .ec-marker--demo,
.earthcell-content .ec-c10 p {
	font-family: Epilogue, sans-serif;
	font-size: 13px;
	line-height: 1.5;
	font-weight: 400;
	letter-spacing: normal;
	text-transform: none;
}

.earthcell-content > .ec-c3 .ec-credit,
.earthcell-content > .ec-c6 .ec-credit,
.earthcell-content > .ec-c6 .ec-c6__status {
	color: #a9cccf;
}

.earthcell-content .ec-sitecards__note,
.earthcell-content .ec-marker--demo,
.earthcell-content .ec-c10 p {
	color: #5d6672;
}

/* Field labels, tags, years and chips. 14px Epilogue, the h6 role. */

.earthcell-content .ec-c8__label,
.earthcell-content .ec-c5__year,
.earthcell-content .ec-c5__status,
.earthcell-content .ec-c5__chip,
.earthcell-content .ec-c10__eyebrow,
.earthcell-content .ec-choose__count,
.earthcell-content .ec-choose__value {
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
}

.earthcell-content .ec-c8__label,
.earthcell-content .ec-c5__chip {
	color: #5d6672;
}

.earthcell-content .ec-c10__eyebrow {
	color: var(--ec-teal);
}

/* The site cards' scale line is the handoff's 14px weight 700 in teal. */

.earthcell-content .ec-sitecard__scale {
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-teal);
}

/* Satellite / Drone / Ground: h3 at the handoff's 25px, lime on teal.

   THE THREE PANELS KEEP THEIR SATELLITE PICTURES, AND THAT IS A DECISION TOO.
   The handoff wants the drone and ground panels to be empty states: a dashed
   box saying "No drone capture yet" and "No field evidence yet", because
   neither capture exists. Asked on 7 September 2026, Piers kept the pictures:
   "let's keep satellite pictures for now, as they help."

   WHAT MAKES THAT SAFE IS THE STATUS LINE UNDER EACH ONE, AND IT MUST STAY.
   The drone panel says "Not yet flown · shown at approximate scale" and the
   ground panel says "No field evidence recorded yet", in the reader's own
   words, directly under the picture. The band admits in writing that two of
   its three pictures are the same satellite imagery at a closer zoom. Delete
   either line and the panels start implying a capture that has not happened,
   which is the fault the handoff was guarding against. The test in
   widget/test/chapters.test.ts guards both lines. */

.earthcell-content > .ec-c6 .ec-c6__kind {
	font-family: Epilogue, sans-serif;
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-lime);
}

/* The enquiry panel's heading, which used to be a marker in capitals. */

.earthcell-content .ec-panel__heading {
	margin: 0 0 12px;
	font-family: Epilogue, sans-serif;
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-ink);
}

/*
 * THE TWO USES THAT SURVIVE, AND THEY ARE NAMED HERE SO THE NEXT SWEEP DOES NOT
 * TAKE THEM TOO. The cell code and the sample status line are machine values.
 * They sit inside .ec-c8__records, which the handoff hides on this page, and
 * the same treatment belongs on the widget's own record panel when it draws
 * one.
 */
.earthcell-content .ec-c8__chip-code,
.earthcell-content .ec-c8__chip-status {
	font-family: var(--ec-mono);
}

/* THE YEAR BUTTONS NEEDED NAMING THE LONG WAY, and this is the third time on
   this page that specificity has beaten source order rather than a rule being
   wrong. The block above says .earthcell-content .ec-c5__year, which is two
   classes. An earlier rule says
   `.earthcell-content > .ec-c5 .ec-c5__years .ec-c5__year[data-ec-year-button]`
   — four classes and an attribute — so the years stayed monospace while every
   other label changed. Read the computed value back off the page after any
   sweep like this; the file looking right proves nothing. */

.earthcell-content > .ec-c5 .ec-c5__years .ec-c5__year[data-ec-year-button] {
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
}

/* ==========================================================================
   THE TYPE SCALE. 20 / 25 / 31 / 39 / 49 / 61, PLUS 18, 16, 14 AND 13.

   The handoff's reason for having one at all: "The theme declares twelve ad-hoc
   sizes (12, 14, 16, 18, 20, 24, 26, 28, 30, 32, 36, 42), which is an
   accumulation rather than a scale."

   MEASURED ON STAGING BEFORE THIS BLOCK. Fifty-three distinct size-and-class
   pairings carried text, and thirteen of them sat on no step of the scale:

     40px   three of the nine section headings, where six were already 39
     27px   the pull line under "Follow what happens next"
     24px   the "Project partners" sub-heading, where the handoff asks for 31
     21px   the three site card names
     19px   the three principles' headings
     17px   the record panel's heading
     15.2px the map fallback sentence
     15px   the record panel's empty line and its button
     12px   the breadcrumb

   Six of those nine headings being 39 and three being 40 is the whole problem
   in miniature: nobody chose 40, it was left over, and one pixel of difference
   between two headings on one page is not a decision anybody can defend.

   THE BREADCRUMB IS LEFT AT 12px. It is the site's own breadcrumb, shared with
   pages this handoff says nothing about, and it is furniture above the article
   rather than part of the page's own type.

   EVERY SELECTOR BELOW IS WRITTEN AT THREE CLASSES OR MORE ON PURPOSE. The
   rules it has to beat are two classes and an element, which outranks two
   classes on its own. That has caught this page four times now: the eyebrows,
   the year buttons, band 08's lede and band 04's heading. Read the computed
   value back off the rendered page after deploying, never off the file.
   ========================================================================== */

/* 39 for every section heading, and the clamp is the handoff's own. */

.earthcell-content > .ec-c3 .ec-c3__title,
.earthcell-content > .ec-c5 .ec-c5__title,
.earthcell-content > .ec-section--rule .ec-col > h2,
.earthcell-content > .ec-section--narrow > h2 {
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	text-wrap: pretty;
}

/* 31 for the one sub-heading on the page. It is a division inside band 05, not
   a band of its own, which is why it is an h3 and not an h2 now: the handoff
   marks that division with 96px of space and this heading rather than with the
   rule that used to sit above it. */

.earthcell-content > .ec-partners .ec-col > .ec-partners__title {
	margin: 0 0 24px;
	font-size: 31px;
	line-height: 1.16;
	font-weight: 700;
}

/* 25 for every h3 and for the pull line. */

.earthcell-content > .ec-section--rule .ec-sitecard .ec-sitecard__name,
.earthcell-content > .ec-c3 .ec-trio .ec-trio__title,
.earthcell-content > .ec-c9 .ec-choose .ec-choose__heading {
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
}

.earthcell-content > .ec-c5 .ec-c5__strapline {
	font-size: 25px;
	line-height: 1.4;
	font-weight: 600;
}

/* 18 for body copy inside the chooser, which was on three sizes of its own. */

.earthcell-content > .ec-c9 .ec-choose .ec-choose__fallback,
.earthcell-content > .ec-c9 .ec-choose .ec-choose__empty p,
.earthcell-content > .ec-c9 .ec-choose .ec-choose__go {
	font-size: 18px;
	line-height: 1.6;
}

/*
 * 16 FOR THE PROTOTYPE NOTE, NOT 13, AND THE REFERENCE SETTLES IT.
 *
 * The handoff's type table puts captions, credits and disclaimers at 13, and an
 * earlier pass in this file read the prototype note as one of those. The
 * rendered reference has it at 16: it is a note a reader is meant to read, not
 * a credit under a picture. Section 4 asks only that it stays "mono-free small
 * type, so it reads as a footnote not a twelfth band", and 16 against the
 * page's 18 body does that.
 */
.earthcell-content > .ec-c10 p {
	font-size: 16px;
	line-height: 1.6;
}

/* --------------------------------------------------------------------------
   THE SITE CARDS, TO THE REFERENCE.

   The scale line moved out of the thumbnail in the page body, so it is now an
   ordinary line of the card rather than a caption over imagery. It takes the
   handoff's label treatment: 14px weight 700 in teal, above the name. See the
   note beside the paper rule further up this file for what it was doing before
   and what it measured.

   The card body goes to 18px, which is the page's body size and the
   reference's. It was 16, one step below every other paragraph on the page for
   no stated reason.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-section--rule .ec-sitecard .ec-sitecard__scale {
	order: 1;
	position: static;
	inset: auto;
	margin: 0;
	padding: 0;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 700;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-teal);
	text-shadow: none;
	background: none;
}

.earthcell-content > .ec-section--rule .ec-sitecard .ec-sitecard__body {
	font-size: 18px;
	line-height: 1.6;
}


/* ==========================================================================
   THE TWO BIOESTRELA HUB PAGES TAKE THE KOGI SITE PATTERN.
   Added 7 September 2026.

   /earthcell/bioestrela-valezim/ and /earthcell/bioestrela-furtado/ were built
   as flat content: no sections at all, every heading and paragraph a direct
   child of the content wrapper, and every rule that made them look like
   anything written as a `style` attribute in the page body. That is the state
   the Kogi page was in before the Site Pattern handoff of 7 September, and the
   fix is the same one: bands.

   WHY THESE PAGES DO NOT SIMPLY REUSE .ec-c3, .ec-c4, .ec-c5 AND THE REST.
   Those classes carry Kogi's grounds in a fixed order AND Kogi's own component
   rules: a satellite tile grid, a timeline of years, three capture-scale
   panels, a hex grid of 143 cells. Bioestrela has none of those and has four
   components Kogi does not. Borrowing the names would put a Bioestrela page one
   edit away from inheriting a rule written for a Kogi component.

   So the ground is named for what it is rather than for its position, and the
   sequence lives in the page body where a reader of that file can see it. The
   Kogi bands should fold into this vocabulary when somebody next has reason to
   touch them; nothing here changes them today.

   THE CLASS IS .ec-hubband AND NOT .ec-ground, WHICH WAS THE FIRST NAME.
   `--ec-ground` is already a custom property on this wrapper holding the night
   colour. A class and a property one character apart is a fault waiting to
   happen in a file this long.

   THE BOUNDARY RULE IS THE HANDOFF'S AND IT IS THE WHOLE POINT: every boundary
   is marked exactly once, by a change of ground, by a photograph, or by a
   hairline. Never two, never none. The sequence each page uses is written at
   the top of its body file, with the boundary marker named for every join. If a
   band is added, removed or reordered, that table is what gets checked.

   FOUR TONES PER GROUND, SO A COMPONENT DOES NOT HAVE TO KNOW WHERE IT SITS.
   --ec-hair, --ec-accent and --ec-quiet are set by the ground and read by every
   component below. The Darnelas callout has a 2px rule down its left edge: teal
   on paper, lime on teal. Without this it would have been teal on teal, which
   is invisible, and the page would have looked finished.
   ========================================================================== */

/* --------------------------------------------------------------------------
   THE BAND, AND THE COLUMN INSIDE IT.

   THE FULL-BLEED SUM IS THE FILE'S OWN, used by .ec-band, by every Kogi band
   and by .ec-photoband, and it is used again here rather than a cleverer one.
   See the long note beside the Kogi band list: an invalid calc does not fall
   back, it reverts the property to 0, and the band then keeps its container's
   left edge and sits hundreds of pixels to the right on a wide window.

   THE CONTENT COLUMN IS A DIV INSIDE THE SECTION AND THAT IS NOT TIDINESS.
   Putting `max-width: var(--ec-band); margin-inline: auto` on every direct
   child centres each child on its own width, so a 690px paragraph sits in the
   middle of a 1040px column while the heading above it starts at the left edge.
   One inner div takes the 1040 and everything inside it is flush left, which is
   what the handoff's alignment table requires: offset 0, the container's left
   edge, carrying every eyebrow, heading and paragraph that opens a block.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-hubband {
	width: 100vw;
	margin: 0 0 0 calc( 50% - 50vw );
	padding: var(--ec-band-pad) var(--ec-gutter);
	border: 0;
}

.earthcell-content > .ec-hubband > .ec-hubband__inner {
	max-width: var(--ec-band);
	margin: 0 auto;
}

/*
 * INHERIT, RATHER THAN A COLOUR PER GROUND PER ELEMENT.
 *
 * The band sets `color` and everything in it should take that. It does not,
 * because single-post.css loads last of the five stylesheets on this page and
 * sets its own ink on headings and paragraphs. `color: inherit` re-asserts the
 * band's colour without this block having to know what that colour is, so a
 * ground added later needs no matching ink rule.
 *
 * IT STOPS TWO LEVELS DOWN ON PURPOSE. The choosing map and the enquiry form
 * are drawn by the widget, several levels deeper, and they carry their own
 * palette. Reaching them would put paper text on a paper card.
 */
.earthcell-content > .ec-hubband > .ec-hubband__inner > :is( h2, h3, p, ul, ol ),
.earthcell-content > .ec-hubband > .ec-hubband__inner > * > :is( h2, h3, p, li ) {
	color: inherit;
}

/* --------------------------------------------------------------------------
   THE FOUR GROUNDS.

   Contrast, from the handoff's measured table: slate on paper 9.1:1, paper on
   teal 7.4:1, slate on lime 7.6:1, paper on night 15.8:1. Never paper on lime,
   which is 1.2:1, which is why the lime ground takes slate ink.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-hubband--paper {
	--ec-hair: rgba( 62, 68, 84, 0.18 );
	--ec-accent: var(--ec-teal);
	--ec-quiet: #5d6672;
	background: var(--ec-paper);
	color: var(--ec-ink);
}

.earthcell-content > .ec-hubband--teal {
	--ec-hair: rgba( 244, 248, 249, 0.24 );
	--ec-accent: var(--ec-lime);
	--ec-quiet: #a9cccf;
	background: var(--ec-teal);
	color: var(--ec-paper);
}

.earthcell-content > .ec-hubband--lime {
	--ec-hair: rgba( 40, 81, 63, 0.28 );
	--ec-accent: #28513f;
	--ec-quiet: #3e4454;
	background: var(--ec-lime);
	color: var(--ec-ink);
}

.earthcell-content > .ec-hubband--night {
	--ec-hair: rgba( 244, 248, 249, 0.2 );
	--ec-accent: var(--ec-lime);
	--ec-quiet: #8fa39a;
	background: #05080b;
	color: var(--ec-paper);
}

/*
 * THE ONE HAIRLINE, AND IT MEANS "SECTION BOUNDARY".
 *
 * It is full bleed, running to both edges like the ground it sits on, because
 * it marks the band edge and not the content. Each page uses it exactly once,
 * at its single paper-to-paper join. Component rules inside a band are a
 * different thing and stay at track width.
 */
.earthcell-content > .ec-hubband--rule {
	border-top: 1px solid rgba( 62, 68, 84, 0.16 );
}

/* --------------------------------------------------------------------------
   TYPE. The handoff's ratio scale, 20 / 25 / 31 / 39, and nothing else.

   A HEADING INSIDE A BAND TAKES NO TOP MARGIN: the band is the space. The
   site's own `.earthcell-content > h2` carries `margin: 88px 0 1.5rem`, which
   these headings escape only because they are no longer direct children. The
   margin is written out here so that stays true if anything changes.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-hubband__inner > h2 {
	margin: 0 0 24px;
	max-width: 486px;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	letter-spacing: normal;
	text-wrap: pretty;
}

.earthcell-content .ec-hubband__inner > p {
	margin: 0 0 16px;
	max-width: 690px;
	font-size: 18px;
	line-height: 1.6;
}

.earthcell-content .ec-hubband__inner > p:last-child {
	margin-bottom: 0;
}

/* The eyebrow takes the ground's accent: teal on paper, lime on teal and on
   night, #28513f on lime. The class is named four times because the rule it has
   to outrank names it three; see the note beside that rule for why it grew. */

.earthcell-content > .ec-hubband .ec-eyebrow.ec-eyebrow.ec-eyebrow.ec-eyebrow {
	margin: 0 0 20px;
	color: var(--ec-accent);
}

/* --------------------------------------------------------------------------
   THE COMPONENTS.

   Every one of these was a `style` attribute in the page body until today.
   WPBakery strips those the first time anybody opens the page in the builder
   and saves, so a page that looks finished today is one save from looking like
   an unstyled document, with nothing in the source to say what it lost.

   THE THREE-UPS SHARE ONE BREAKPOINT. `repeat(auto-fit, minmax(200px, 1fr))`
   for the principles, the organisations, the facts and the years, so they fold
   to one column together. Three different floors is how one band ends up
   showing three columns while the next shows two at the same width.
   -------------------------------------------------------------------------- */

/* The strip of project facts. Label over value, which is the site's own
   case-study pattern and the handoff's stat row. */

.earthcell-content .ec-hubband .ec-facts {
	margin: 32px 0 48px;
	padding-top: 24px;
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 200px, 1fr ) );
	gap: 24px;
	border-top: 1px solid var(--ec-hair);
}

.earthcell-content .ec-hubband .ec-facts__label {
	margin: 0 0 8px;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 700;
	color: var(--ec-quiet);
}

.earthcell-content .ec-hubband .ec-facts__value {
	font-size: 39px;
	line-height: 1.05;
	font-weight: 800;
}

/* The two fire years. The year is the figure and the approved sentence sits
   under it. */

.earthcell-content .ec-hubband .ec-years {
	margin: 32px 0 48px;
	padding-top: 24px;
	max-width: 690px;
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 200px, 1fr ) );
	gap: 24px;
	border-top: 1px solid var(--ec-hair);
}

.earthcell-content .ec-hubband .ec-years__value {
	margin: 0 0 12px;
	font-size: 31px;
	line-height: 1.16;
	font-weight: 700;
}

.earthcell-content .ec-hubband .ec-years__note {
	margin: 0;
	font-size: 18px;
	line-height: 1.6;
}

/* The Darnelas callout: the one thing on the Valezim page that has already
   happened in the ground. The rule down its left edge takes the ground's
   accent, so it is teal on paper and lime on teal rather than invisible. */

.earthcell-content .ec-hubband .ec-callout {
	margin: 32px 0 48px;
	max-width: 690px;
	padding: 4px 0 4px 28px;
	border-left: 2px solid var(--ec-accent);
}

.earthcell-content .ec-hubband .ec-callout__value {
	margin: 0 0 12px;
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
}

.earthcell-content .ec-hubband .ec-callout__note {
	margin: 0;
	font-size: 18px;
	line-height: 1.6;
}

/* The three-up used twice on each page: the principles, and the organisations. */

.earthcell-content .ec-hubband .ec-cards {
	margin: 48px 0 0;
	padding-top: 24px;
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 200px, 1fr ) );
	gap: 24px;
	border-top: 1px solid var(--ec-hair);
}

.earthcell-content .ec-hubband .ec-cards__item > h3 {
	margin: 0 0 12px;
	max-width: none;
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: normal;
}

.earthcell-content .ec-hubband .ec-cards__item > p {
	margin: 0;
	font-size: 18px;
	line-height: 1.6;
}

/* What an EarthCell is not. Four approved words, set as four lines, which is
   the page's clearest disclaimer. */

.earthcell-content .ec-hubband .ec-not {
	margin: 32px 0 32px;
	padding-top: 24px;
	max-width: 588px;
	border-top: 1px solid var(--ec-hair);
}

.earthcell-content .ec-hubband .ec-not__list {
	margin: 0;
	padding: 0;
	max-width: none;
	list-style: none;
}

/* THE CLASS IS NAMED FOUR TIMES, AND THE RULE IT HAS TO BEAT IS NOT IN THIS
   FILE AT ALL. The parent theme's style.css carries

     .page .entry-content ul:not(.category-filters):not(.clevel-subnav):not(.flight-details) li

   which scores five classes to an element, because each `:not()` argument
   counts as the class inside it. This rule at three classes lost on font-size
   and line-height while its padding and its border applied, so the four
   disclaimer lines rendered at 20px on a 32px line inside a component that was
   otherwise correct. Four names put it at six classes, above that rule whatever
   order the two stylesheets load in.

   MEASURED, NOT ASSUMED: read the computed value back off the deployed page.
   Every fault found on this page today was a valid rule that matched and still
   lost. */

.earthcell-content .ec-hubband .ec-not__list.ec-not__list.ec-not__list.ec-not__list > li {
	margin: 0;
	padding: 12px 0;
	border-bottom: 1px solid var(--ec-hair);
	font-size: 18px;
	line-height: 1.5;

	/* THE SAME RULE ALSO MAKES EVERY LIST ITEM A FLEX ROW WITH AN EMPTY 24px
	   BULLET SLOT IN FRONT OF IT. `list-style: none` does not touch that: the
	   slot is a ::before with `content: ""`, not a marker. Measured on the
	   deployed page, the four lines started at x=476 while the heading above
	   them started at 440, which is the alignment line every other block on the
	   page sits on. */
	display: block;
}

.earthcell-content .ec-hubband .ec-not__list.ec-not__list.ec-not__list.ec-not__list > li::before {
	content: none;
}

.earthcell-content .ec-hubband .ec-not__list > li:first-child {
	padding-top: 0;
}

.earthcell-content .ec-hubband .ec-not__list > li:last-child {
	padding-bottom: 0;
	border-bottom: 0;
}

/* The pull line under the opening prose. */

.earthcell-content .ec-hubband .ec-lead {
	margin: 32px 0 0;
	max-width: 690px;
	font-size: 25px;
	line-height: 1.4;
	font-weight: 600;
}

/* --------------------------------------------------------------------------
   CONTAINED FIGURES.

   The two photographs that separate bands are full-bleed .ec-photoband
   sections and use the Kogi page's own rules. These are the ones that stay
   inside a band because they illustrate the words beside them rather than
   marking a boundary. A contained figure is not a boundary marker and must not
   sit at a band edge, or the join would carry two markers.

   THE CAPTION IS 13px IN THE GROUND'S QUIET TONE, never uppercase and never
   monospace. The handoff allows monospace in two places on the whole design and
   a caption is neither of them.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-hubband .ec-figure {
	margin: 48px 0 0;
	max-width: 690px;
}

.earthcell-content .ec-hubband .ec-figure > img {
	display: block;
	width: 100%;
	height: auto;
	border-radius: 10px;
}

.earthcell-content .ec-hubband .ec-figure__caption {
	margin: 12px 0 0;

	/* THE TYPEFACE HAS TO BE NAMED, AND HERE IS WHAT HAPPENS WITHOUT IT.
	   `.earthcell-content .ec-figure__caption` further up this file sets
	   `font-family: var(--ec-mono)`, from the draft where captions were machine
	   labels. This rule reset the size, the colour, the letter-spacing and the
	   capitals but not the family, so both captions rendered at 13px in
	   monospace: everything about them corrected except the one thing anybody
	   would notice. The handoff allows monospace in two places on the whole
	   design, the cell code and its sample marker, and a caption is neither. */
	font-family: inherit;

	font-size: 13px;
	line-height: 1.5;
	font-weight: 400;
	letter-spacing: normal;
	text-transform: none;
	color: var(--ec-quiet);
}

/* The one tall photograph on each page sets its caption beside it rather than
   under it, so a portrait does not push the band that follows a long way down.
   It folds to one column on the same 200px floor as every other three-up. */

.earthcell-content .ec-hubband .ec-figure--portrait {
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 200px, 1fr ) );
	gap: 24px;
	align-items: end;
}

.earthcell-content .ec-hubband .ec-figure--portrait > .ec-figure__caption {
	margin: 0 0 4px;
}

/* --------------------------------------------------------------------------
   THE CLOSING BAND: THE ASK, ON A PHOTOGRAPH.

   This is the Kogi page's band 10 and it is what marks the boundary above it.
   The section keeps the id `#enquiry`, because that is the anchor a reader
   follows to write to us, and the existing `#enquiry` rules already lay it out
   as words left and the form card right. Three things change.

   THE 104px TOP MARGIN GOES. Bands butt directly and space each other with
   padding. A margin between two coloured grounds paints the page ground through
   the join, which is a third thing at the boundary and reads as a gap rather
   than an edge.

   THE SCRIM RUNS LEFT TO RIGHT, not top to bottom. A top-down scrim darkens the
   middle of the picture, which is the part a reader is meant to look at. Here
   it also has to hold the words beside the form legible on a photograph that is
   bright at the top.

   THE GROUND UNDER THE PHOTOGRAPH IS #1c2c26, the handoff's own value for this
   band, so it still reads as a band if the image never arrives.

   EVERY CHILD IS RAISED ABOVE THE SCRIM. The scrim is a positioned pseudo
   element at z-index 1, so anything left at `auto` sits behind it and
   disappears. This is the whole band's content, so it is not a detail.
   -------------------------------------------------------------------------- */

.earthcell-content > #enquiry.ec-hubband--photo {
	--ec-hair: rgba( 244, 248, 249, 0.24 );
	--ec-accent: var(--ec-lime);
	--ec-quiet: #a9bcc0;
	position: relative;
	overflow: hidden;
	margin-top: 0;
	background: #1c2c26;
}

.earthcell-content > #enquiry.ec-hubband--photo > .ec-enquiry-photo {
	position: absolute;
	inset: 0;
	z-index: 0;
	display: block;
	width: 100%;
	height: 100%;
	max-width: none;
	margin: 0;
	object-fit: cover;
}

.earthcell-content > #enquiry.ec-hubband--photo::before {
	content: "";
	position: absolute;
	inset: 0;
	z-index: 1;
	background: linear-gradient( 90deg, rgba( 5, 20, 15, 0.94 ) 0%, rgba( 5, 20, 15, 0.86 ) 45%, rgba( 5, 20, 15, 0.4 ) 80%, rgba( 5, 20, 15, 0.2 ) 100% );
}

.earthcell-content > #enquiry.ec-hubband--photo > .ec-enquiry-head,
.earthcell-content > #enquiry.ec-hubband--photo > p,
.earthcell-content > #enquiry.ec-hubband--photo > .ec-enquiry {
	position: relative;
	z-index: 2;
}

/* The label and the heading travel as one block spanning both columns. The
   theme pins the form to `grid-area: 2 / 2 / span 99` and tells the heading to
   span both columns, which only works while the heading is the first item. A
   label added as a direct child takes row one, column one, and the heading and
   every paragraph fall below the form, leaving the left column empty. Piers saw
   exactly that on 4 September 2026: "There's a contact form on the right, but
   nothing on the left." */

.earthcell-content > #enquiry > .ec-enquiry-head {
	grid-column: 1 / -1;
	max-width: none;
}

.earthcell-content > #enquiry > .ec-enquiry-head > h2 {
	margin: 0;
	max-width: 486px;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	letter-spacing: normal;
	color: var(--ec-paper);
	text-wrap: pretty;
}

.earthcell-content > #enquiry > .ec-enquiry-head > .ec-eyebrow.ec-eyebrow.ec-eyebrow.ec-eyebrow {
	margin: 0 0 20px;
	color: var(--ec-lime);
}

/* The prose beside the form takes the handoff's 20px standfirst, on the same
   588px line as the form card. `46ch` was the old measure and it moves with the
   font size, so no two blocks on the page shared a right edge. */

.earthcell-content > #enquiry.ec-hubband--photo > p {
	max-width: 588px;
	font-size: 20px;
	line-height: 1.6;
	color: #dbe6e8;
}

/* --------------------------------------------------------------------------
   THE ONE HEADING THE HUBBAND BLOCK ABOVE DOES NOT NAME.

   "What an EarthCell is not" was an eyebrow above the list of four, which is a
   label doing a heading's job at label size. It is an h3 now, and it needs the
   scale's 25px like every other h3 in a band: without this it falls through to
   the site's own h3, which is a different size and carries a top margin.

   THE LIST ITSELF IS `.ec-not__list` AND ITS RULES ARE ABOVE. This adds the
   heading and nothing else.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-hubband .ec-not > h3 {
	margin: 0 0 16px;
	max-width: none;
	font-size: 25px;
	line-height: 1.2;
	font-weight: 700;
	letter-spacing: normal;
}

/* --------------------------------------------------------------------------
   A CONTAINED FIGURE IS A BLOCK, SO WHATEVER FOLLOWS IT NEEDS THE BLOCK GAP.

   `.ec-hubband__inner > p` carries its space underneath (`margin: 0 0 16px`)
   rather than above, which is right for a run of paragraphs and wrong directly
   after a figure or a three-up: the paragraph butts against the picture with
   16px of its own space hanging below it instead.

   Both of the Bioestrela pages end a band this way at least once.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-hubband__inner > .ec-figure + p,
.earthcell-content .ec-hubband__inner > .ec-cards + p,
.earthcell-content .ec-hubband__inner > .ec-choose + p {
	margin-top: 48px;
}

/* --------------------------------------------------------------------------
   THE EMPTY PARAGRAPHS WORDPRESS LEAVES INSIDE A BAND.

   `wpautop` wraps a blank line between two block elements in a paragraph of its
   own, and sometimes in a paragraph holding a single `<br>`. Measured on the
   rebuilt Furtado page: eighteen of them, ten of which sit inside a band. Each
   one carries the band's paragraph margin, so a band picks up 16px of space it
   was never given, unevenly, wherever the source file happens to have a blank
   line in it.

   THE THEME ALREADY DOES THIS AT THE TOP LEVEL and has since the page was flat:
   `.earthcell-content > p:empty`. The paragraphs inside a section are no longer
   direct children, so that rule stopped reaching them the moment these pages
   got sections. This is the same rule, one level in.

   `:empty` DOES NOT MATCH A PARAGRAPH HOLDING A `<br>`, because a br is a child.
   The `:has` selector beside it catches those. Both were present on the page.

   THE SOURCE FILE COULD AVOID THIS by having no blank lines between block
   elements, and the body files say so where it matters. It is not a rule a
   person can be relied on to hold across two long files, so the stylesheet
   carries it instead.
   -------------------------------------------------------------------------- */

.earthcell-content .ec-hubband p:empty,
.earthcell-content .ec-hubband p:has( > br:only-child ),
.earthcell-content .ec-photoband p:empty,
.earthcell-content > #enquiry p:empty,
.earthcell-content > #enquiry p:has( > br:only-child ) {
	display: none;
}

/* --------------------------------------------------------------------------
   A BAND THAT IS ONLY PROSE PUTS ITS HEADING IN A COLUMN OF ITS OWN.

   This is the handoff's first row shape, 5 columns of heading and 7 of prose,
   and it is the rule the Kogi page already follows. Measured on staging at
   1920px: Kogi's band 02, which is nothing but a heading and three paragraphs,
   runs the heading at x=432 w=420 and the prose at x=876 w=596. Every other
   Kogi band runs both flush left at 432, because every other Kogi band has a
   component under the prose that fills the container.

   SO THE RULE IS ABOUT WHAT IS IN THE BAND, NOT WHERE THE BAND SITS. A band
   holding cards, a map, a list or a figure keeps one left edge, because the
   component and the prose have to share it. A band holding only paragraphs has
   nothing to fill its right third, and 350px of empty paper beside three
   paragraphs reads as a page that stopped early.

   THE FIRST PARAGRAPH IS 20px, WHICH IS THE HANDOFF'S STANDFIRST. Kogi's band
   02 does the same. It is the only place on either page where a paragraph is
   not 18px.

   THE TRACK WIDTH GOES IN THE FLEX BASIS AND NEVER IN flex-grow OFF A ZERO
   BASIS. With flex-basis 0 a padded track's padding lands outside its grown
   share, so any padded panel ends up wider than its proportion. Percentage
   bases include padding under border-box, so the two bases sum to the space and
   the ratio holds whatever the padding is. The handoff measured that fault at
   1.203 and 1.331 against a target of 1.420.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-hubband--split > .ec-hubband__inner {
	display: flex;
	flex-wrap: wrap;
	gap: var(--ec-gutter);
	align-items: start;
}

.earthcell-content > .ec-hubband--split .ec-hubband__words {
	flex: 1 1 calc( ( 100% - 24px ) * 486 / 1176 );
	min-width: min( 300px, 100% );
}

.earthcell-content > .ec-hubband--split .ec-hubband__prose {
	flex: 1 1 calc( ( 100% - 24px ) * 690 / 1176 );
	min-width: min( 340px, 100% );
}

/* The heading fills its own track rather than keeping the 486px cap, which was
   the whole container's heading line and is now the track itself. */

.earthcell-content .ec-hubband__words > h2 {
	margin: 0;
	max-width: none;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	letter-spacing: normal;
	text-wrap: pretty;
}

.earthcell-content .ec-hubband__prose > p {
	margin: 0 0 16px;
	max-width: none;
	font-size: 18px;
	line-height: 1.6;
}

.earthcell-content .ec-hubband__prose > p:first-child {
	font-size: 20px;
}

.earthcell-content .ec-hubband__prose > p:last-child {
	margin-bottom: 0;
}

/*
 * A SPLIT BAND'S COLUMN HOLDS TWO TRACKS AND NOTHING ELSE.
 *
 * Measured on the Furtado page after the split went on: the flex row had four
 * items, not two. `wpautop` had put a bare `<br>` and a paragraph wrapping an
 * HTML comment in front of the two tracks. Both are invisible and both take a
 * flex item, and between them plus three gaps they pushed the prose track past
 * 100% of the row, so it wrapped to a second line and the heading track grew to
 * fill the first. The heading rendered 1016px wide at an offset of 456 instead
 * of 419 at 432, which looks like a broken ratio rather than an extra child.
 *
 * The two tracks are the only things a split band's column is ever meant to
 * hold, so saying that is both the fix and the check.
 */
.earthcell-content > .ec-hubband--split > .ec-hubband__inner > :not( .ec-hubband__words ):not( .ec-hubband__prose ) {
	display: none;
}

/* --------------------------------------------------------------------------
   AND THE BARE `<br>`, WHICH IS THE SAME FAULT WEARING A DIFFERENT TAG.

   The rule above hides the empty paragraphs wpautop leaves behind, and a
   paragraph holding nothing but a `<br>`. It does not catch a `<br>` that is a
   direct child of the section, which is what wpautop writes when a blank line
   follows an element it will not wrap, such as the closing band's photograph.

   IN A GRID EACH ONE STEALS A CELL, and that is not a cosmetic 16px. Measured
   on the rebuilt Valezim page: `#enquiry` is a two-column grid whose form is
   pinned to `grid-row: 2 / span 99` and whose heading block spans both columns.
   A stray `<br>` took row one, so the heading and all three paragraphs dropped
   to row three and the form card floated 750px above the words it belongs to.
   Nothing in the page source said why.

   The Kogi bands already carry this rule; see `.ec-c1 > br` and its list
   further up the file. These are the bands that did not exist when it was
   written.
   -------------------------------------------------------------------------- */

.earthcell-content > .ec-hubband > br,
.earthcell-content > .ec-hubband > .ec-hubband__inner > br,
.earthcell-content > .ec-photoband > br,
.earthcell-content > #enquiry > br {
	display: none;
}


/* ==========================================================================
   MOBILE PASS ON THE EARTHCELL SECTION, 7 September 2026.

   Piers, looking at /earthcell/what-is-an-earthcell/ on a phone: "all out of
   kilter". He was pointing at the share row, and that fix is in
   css/single-post.css, because THIS FILE DOES NOT LOAD ON THAT PAGE.

   THE SECTION RUNS ON THREE TEMPLATES AND FOUR STYLESHEETS, and that is worth
   knowing before editing any of them:
     /earthcell/                 template-calculator-hub.php  calculator-hub.css
     the five explaining pages   template-article.php         earthcell-section.css
                                                              + single-post.css
     the three project pages     template-earthcell.php       earthcell.css
   Only earthcell-tokens.css and style.css are on all of them. A rule written
   here reaches the three project pages and nothing else.
   ========================================================================== */

/* --------------------------------------------------------------------------
   THE CHOOSING PANEL SET ITS SENTENCES AT 13 AND 14px ON A PHONE.

   The panel is the widget's, and its type was sized for a 460px card sitting
   beside a map on a desktop. On a phone the card is the full column and the
   same sizes are simply small: measured at 320 to 430px, five strings under
   16px, including the one that matters most.

   `.ec-choose__off` at 13px is the sentence that says contributions are not
   open yet and to tell us what you have in mind. It is the whole call to action
   of the panel on a page that cannot take money, and it was the smallest text
   in it.

   THE MAP KEEPS ITS OWN SIZES. Only the panel beside it changes, because the
   map's labels are drawn to a scale that has to match the ground.

   THE BREAKPOINT IS 900 AND NOT 730. At 730 the fix missed 768, which is one of
   the widths this was checked at, and the panel is already full width there.
   -------------------------------------------------------------------------- */

@media screen and ( max-width: 900px ) {

	.earthcell-content .ec-choose__panel .ec-choose__off,
	.earthcell-content .ec-choose__panel .ec-choose__row,
	.earthcell-content .ec-choose__panel .ec-choose__summary,
	.earthcell-content .ec-choose__panel .ec-choose__tell {
		font-size: 16px;
		line-height: 1.5;
	}

	.earthcell-content .ec-choose__fallback {
		font-size: 16px;
		line-height: 1.5;
	}

	/* The one control in the panel, sized for a finger. */
	.earthcell-content .ec-choose__panel .ec-choose__go {
		min-height: 44px;
	}
}

/*
 * THREE FIXES FROM THE VISUAL PASS OF 11 SEPTEMBER 2026, after Piers's replacement Kogi copy.
 *
 * A PAPER BAND ONLY RUNS FLUSH WHEN THE PARTNERS BAND CONTINUES IT. `.ec-section--rule` has no
 * bottom padding because the programme section runs straight into the partners band on the same
 * ground. The new "cannot wait" section uses the same class and is followed by the dark map
 * chapter, so its last line sat on the edge of the next band. Piers found it. This gives the
 * padding back to any rule section that is not followed by the partners band.
 */
.earthcell-content > .ec-section--rule:not( :has( + .ec-partners ) ) {
	padding-bottom: var(--ec-band-pad);
}

/*
 * THE PARTNERS HEADING IS A SECTION HEADING NOW. It was an h3 subtitle at 31px under the
 * programme heading. Piers's copy makes it a section of its own, "Kogi authority and trusted
 * project relationships come first", so it takes the same size as the other section headings.
 */
.earthcell-content > .ec-partners .ec-col > h2.ec-partners__title {
	max-width: 22ch;
	font-size: clamp( 28px, 3.4vw, 39px );
	line-height: 1.14;
	font-weight: 800;
	text-wrap: pretty;
}

/*
 * A LINK IN THE DARK MAP CHAPTER WAS DARK TEAL ON BLACK and could barely be seen. It takes the
 * same lime the other two dark chapters already give their links.
 *
 * THE PLAIN SELECTOR LOST, FOUND ON THE SECOND PASS. style.css colours every unclassed link in a
 * paragraph with `.entry-content p:not(.learn-more) a:not([class])`, which scores three classes
 * and two elements and beat `.ec-c9 a:not(.ec-btn)` at three classes and one. Naming the
 * paragraph as well puts these ahead. The same rule covers the two teal chapters, whose own link
 * rule has the same weakness and whose new "Read how..." links are plain paragraph links too.
 */
.earthcell-content > .ec-c9 a:not( .ec-btn ),
.earthcell-content > .ec-c9 .ec-c9__inner p a:not( .ec-btn ),
.earthcell-content > .ec-c3 .ec-c3__words p a:not( .ec-btn ),
.earthcell-content > .ec-c6 p.ec-body a:not( .ec-btn ) {
	color: var(--ec-lime);
}

/*
 * THE EVIDENCE LISTS READ AT THE PARAGRAPH SIZE. Piers's Kogi copy put two short lists in the
 * evidence chapter. They picked up the site's generic list size and printed larger than the
 * paragraphs around them, which Piers flags as a font discrepancy wherever it happens.
 */
/*
 * THE FIRST VERSION OF THIS RULE LOST. style.css sizes every page list with
 * `.page .entry-content ul:not(...):not(...):not(...) li` at 20px, five classes and two elements.
 * The Kogi body wrapper carries both .entry-content and .earthcell-content, so naming both, the
 * chapter, its inner block and the list box puts this ahead.
 */
.page .entry-content.earthcell-content > .ec-c5 .ec-c5__inner .ec-c5__pending ul li {
	font-size: 17px;
	line-height: 1.6;
}

/* --------------------------------------------------------------------------
   THE TEXT BLOCK PATTERN. ONE SET OF RULES FOR EVERY TEXT BLOCK ON THE PAGE. Piers, 7 October 2026:
   "consistent design, not something different every block. Use patterns that repeat."

     .ec-hubband--grid      the block
       .ec-eyebrow, h2      full width of the 1040px column; heading on one line where it fits
       .ec-intro            the text, left, in the 690px reading measure
         p.ec-key           the sentence the block turns on: bold, body size
       side item, right     one of: .ec-figure (a photograph) or .ec-side (a shaded fact panel)
       .ec-pair             optional: equal columns below a hairline, one per .ec-story
       .ec-choose           optional: the selection map, full width under the text

   ONLY THE THEME'S FIVE SIZES: label, heading, lead (story headings), body, label (captions).
   Words never change to fit the pattern; elements are only grouped.
   -------------------------------------------------------------------------- */
.earthcell-content .ec-hubband--grid .ec-hubband__inner {
	display: grid; grid-template-columns: minmax(0, 1fr) 340px; column-gap: 64px; align-items: start;
}
.earthcell-content .ec-hubband--grid .ec-eyebrow { grid-column: 1 / -1; margin: 0 0 16px; }
.earthcell-content .ec-hubband--grid h2 {
	grid-column: 1 / -1; max-width: none; font-size: var(--fs-heading); line-height: 1.15; margin: 0 0 32px;
}
.earthcell-content .ec-hubband--grid .ec-intro { grid-column: 1; grid-row: 3; }
.earthcell-content .ec-hubband--grid .ec-intro p { max-width: var(--ec-measure); font-size: var(--fs-body); margin: 0 0 16px; }
.earthcell-content .ec-hubband--grid .ec-intro p.ec-key { font-weight: 700; }
.earthcell-content .ec-hubband--grid :is(.ec-figure, .ec-side) { grid-column: 2; grid-row: 3; margin: 6px 0 0; }
.earthcell-content .ec-hubband--grid .ec-figure img { width: 100%; height: auto; aspect-ratio: 1 / 1; object-fit: cover; border-radius: 6px; }
.earthcell-content .ec-hubband--grid .ec-figure__caption { font-size: var(--fs-label); margin-top: 8px; }
.earthcell-content .ec-hubband--grid .ec-side {
	padding: 24px 26px; border-left: 4px solid var(--ec-teal, #005a69); background: var(--ec-tint, #e2ecee);
	font-size: var(--fs-body); line-height: 1.55; font-weight: 600;
}
.earthcell-content .ec-hubband--grid .ec-pair {
	grid-column: 1 / -1; display: grid; grid-auto-flow: column; grid-auto-columns: minmax(0, 1fr); gap: 64px;
	margin-top: 48px; padding-top: 40px; border-top: 1px solid var(--ec-hair, rgba(62, 68, 84, .18));
}
.earthcell-content :is(.ec-hubband--teal, .ec-hubband--night).ec-hubband--grid .ec-pair { border-top-color: rgba(255, 255, 255, .18); }
/* The stories row takes as many equal columns as it has stories: two in "The work", three partners. */
.earthcell-content .ec-hubband--grid .ec-pair:has(> .ec-story:nth-child(3)) { gap: 48px; }
/* A full-width component under the text, such as the selection map in "Explore". */
.earthcell-content .ec-hubband--grid .ec-choose { grid-column: 1 / -1; margin-top: 16px; }
.earthcell-content .ec-hubband--grid .ec-intro p a { font-weight: 700; }
.earthcell-content .ec-hubband--grid .ec-story > *:first-child { margin-top: 0; }
.earthcell-content .ec-hubband--grid .ec-callout { margin: 0; padding: 0; border: 0; }
.earthcell-content .ec-hubband--grid .ec-story h3 { font-size: var(--fs-lead); line-height: 1.25; margin: 0 0 14px; }
.earthcell-content .ec-hubband--grid .ec-story p { font-size: var(--fs-body); max-width: none; margin: 0 0 16px; }
@media ( max-width: 900px ) {
	.earthcell-content .ec-hubband--grid .ec-hubband__inner, .earthcell-content .ec-hubband--grid .ec-pair { display: block; }
	.earthcell-content .ec-hubband--grid :is(.ec-figure, .ec-side) { margin: 28px 0; }
	.earthcell-content .ec-hubband--grid .ec-figure img { aspect-ratio: 3 / 2; }
	.earthcell-content .ec-hubband--grid .ec-story + .ec-story { margin-top: 32px; }
}

/* THE BREADCRUMB ON A PROJECT PAGE SITS ON THE COLUMN'S LEFT EDGE, AT THE TOP OF THE FIRST BLOCK.
   Piers, 7 October 2026: it was indented 24px from every block below it and floated in a gap of
   its own between the map and "The place". */
/* The page body is the 1040px column plus a 24px gutter inside it, while each block's column is the
   full 1040px. So on a wide screen the breadcrumb steps out by the gutter to share the blocks' edge;
   on a narrow one both sit on the gutter already. Measured: 454 against 430 before, 430 after. */
body .entry-content.earthcell-content > nav.ec-crumb { max-width: none; margin: 0; padding: 28px 0 0; }
@media ( min-width: 1088px ) {
	body .entry-content.earthcell-content > nav.ec-crumb { margin-left: calc(-1 * var(--ec-gutter, 24px)); }
}
/* A hidden comment paragraph sits between the two, so the first block is found by type. */
body .entry-content.earthcell-content > nav.ec-crumb ~ section.ec-hubband--grid:first-of-type { padding-top: 32px; }

/* --------------------------------------------------------------------------
   THE CLOSING BAND'S HEADING AND LINK. 23 October 2026, on La Esmeralda and
   Furtado. Two faults Piers asked to be checked carefully:
     - the heading sat flush on the first paragraph. Every other band heading
       has 24px below it (`.ec-hubband__inner > h2`); the closing band's heading
       lives in `.ec-enquiry-head` with margin 0, so the block takes the gap.
     - the "Discuss supporting ..." link was the site's dark teal on a scrim
       that is 94% near-black: it could barely be read. It takes the band's
       own accent, lime, underlined, as the eyebrow above it already does.
   -------------------------------------------------------------------------- */
.earthcell-content > #enquiry > .ec-enquiry-head {
	margin-bottom: 24px;
}

.earthcell-content > #enquiry.ec-hubband--photo > p a {
	color: var(--ec-lime);
	text-decoration: underline;
	text-underline-offset: 3px;
}

/* --------------------------------------------------------------------------
   A BAND HEADING RUNS THE WIDTH OF ITS COLUMN. 23 October 2026. Piers, on La
   Esmeralda: "why is this h tag on 4 lines". `.ec-hubband__inner > h2` caps
   every band heading at 486px, so "La Esmeralda is a sacred Kogi site in the
   Sierra Nevada de Santa Marta" broke into four short lines inside a 1040px
   column. The text block pattern (`.ec-hubband--grid`, 7 October) already lets
   the heading run the full column, "on one line where it fits"; this does the
   same for every full-width band. Split bands keep their heading column: there
   the heading sits beside the prose and the column is the limit.
   -------------------------------------------------------------------------- */
.earthcell-content .ec-hubband:not(.ec-hubband--split) .ec-hubband__inner > h2 {
	max-width: none;
}

/* --------------------------------------------------------------------------
   NO EMPTY BLOCK ABOVE THE FOOTER. 23 October 2026. Piers: "remove empty
   white block above footer on earthcell pages". `.earthcell-content` keeps
   96px of padding under its last child, for pages that end in plain prose.
   A project page ends in the full-bleed enquiry band, so that padding showed
   as a bare white strip between the photograph and the footer. When the
   enquiry band is there, the page ends where the band ends.
   -------------------------------------------------------------------------- */
.earthcell-content:has(> #enquiry) {
	padding-bottom: 0;
}

/* --------------------------------------------------------------------------
   THE RECORD FROM SPACE. 28 October 2026, on the map-first test page
   (/earthcell/bioestrela-valezim-map-first/ on staging). Piers: the pages
   "lack a sense of real climate tech". The band sits on night under the map,
   so the map and its readings read as one instrument before the story.

     .ec-readout            the band (with .ec-hubband--night)
       .ec-readings         five dated readings in one row, oldest first
         .ec-reading        when (mono label), value, one-line note, source link
         --alert            the fire readings: the value takes the burn orange
       .ec-compare          before-and-after slider. The range input sets
                            --ec-split on the figure; nothing else runs.

   Unused on every other page, so it is safe on live before any page uses it.
   -------------------------------------------------------------------------- */
.earthcell-content > .ec-readout { padding-top: 64px; }
.earthcell-content > .ec-readout .ec-hubband__inner > h2 { margin: 0 0 16px; font-size: var(--fs-heading); }
.earthcell-content > .ec-readout .ec-hubband__inner > p.ec-readout__lead {
	max-width: var(--ec-measure, 690px); margin: 0 0 40px; font-size: var(--fs-body); color: rgba( 244, 248, 249, 0.78 );
}
.earthcell-content .ec-readout .ec-readings {
	/* As many equal columns as readings: five on Valezim and Furtado, three on La Esmeralda and Kogi. */
	display: grid; grid-auto-flow: column; grid-auto-columns: minmax( 0, 1fr ); gap: 0;
	border-top: 1px solid var(--ec-hair); border-bottom: 1px solid var(--ec-hair);
}
.earthcell-content .ec-readout .ec-reading { padding: 24px 24px 26px; display: flex; flex-direction: column; gap: 10px; }
.earthcell-content .ec-readout .ec-reading + .ec-reading { border-left: 1px solid var(--ec-hair); }
.earthcell-content .ec-readout .ec-reading:first-child { padding-left: 0; }
.earthcell-content .ec-readout .ec-reading > p { margin: 0; max-width: none; color: inherit; }
.earthcell-content .ec-readout .ec-reading__when {
	font-family: var(--ec-mono); font-size: 11px; font-weight: 600; line-height: 1.3; letter-spacing: 0.14em;
	text-transform: uppercase; color: var(--ec-aqua);
}
.earthcell-content .ec-readout .ec-reading__value {
	font-size: clamp( 28px, 2.6vw, 39px ); line-height: 1.05; font-weight: 800; color: var(--ec-paper);
	font-variant-numeric: tabular-nums;
}
.earthcell-content .ec-readout .ec-reading--alert .ec-reading__value { color: #ff9a52; }
.earthcell-content .ec-readout .ec-reading__note { font-size: 15px; line-height: 1.45; color: rgba( 244, 248, 249, 0.78 ); }
.earthcell-content .ec-readout .ec-reading__source { margin-top: auto !important; padding-top: 6px; font-family: var(--ec-mono); font-size: 11px; line-height: 1.4; }
.earthcell-content .ec-readout .ec-reading__source a { color: var(--ec-quiet); text-decoration: underline; text-underline-offset: 3px; }
.earthcell-content .ec-readout .ec-reading__source a:hover,
.earthcell-content .ec-readout .ec-reading__source a:focus-visible { color: var(--ec-lime); }

.earthcell-content .ec-readout .ec-compare {
	--ec-split: 50%;
	position: relative; margin: 48px 0 0; aspect-ratio: 2 / 1; overflow: hidden; border-radius: 4px; background: #111;
	touch-action: pan-y;
}
.earthcell-content .ec-readout .ec-compare__img {
	position: absolute; inset: 0; width: 100%; height: 100%; max-width: none; object-fit: cover; margin: 0;
}
.earthcell-content .ec-readout .ec-compare__img--after { clip-path: inset( 0 0 0 var(--ec-split) ); }
.earthcell-content .ec-readout .ec-compare__handle {
	position: absolute; top: 0; bottom: 0; left: var(--ec-split); width: 2px; margin-left: -1px;
	background: var(--ec-lime); pointer-events: none;
}
.earthcell-content .ec-readout .ec-compare__handle::after {
	content: ""; position: absolute; top: 50%; left: 50%; width: 40px; height: 40px; margin: -20px 0 0 -20px;
	border-radius: 50%; background: var(--ec-lime);
	box-shadow: 0 2px 10px rgba( 0, 0, 0, 0.45 );
}
.earthcell-content .ec-readout .ec-compare__handle::before {
	content: "\2039 \203A"; position: absolute; top: 50%; left: 50%; z-index: 1; transform: translate( -50%, -52% );
	font: 700 20px/1 var(--ec-mono); color: #05080b; white-space: nowrap;
}
.earthcell-content .ec-readout .ec-compare__tag {
	position: absolute; top: 16px; padding: 6px 10px; background: rgba( 5, 8, 11, 0.78 );
	font-family: var(--ec-mono); font-size: 11px; font-weight: 600; letter-spacing: 0.14em; text-transform: uppercase;
	color: var(--ec-paper); pointer-events: none;
}
.earthcell-content .ec-readout .ec-compare__tag--before { left: 16px; }
.earthcell-content .ec-readout .ec-compare__tag--after { right: 16px; color: #ff9a52; }
/* A forest map on the right, not a fire: the tag takes the band's lime, not the burn orange. */
.earthcell-content .ec-readout .ec-compare__tag--map { color: var(--ec-lime); }
/* The colour key in the caption, matching the map's two classes (esmeralda.py). */
.earthcell-content .ec-readout .ec-swatch {
	display: inline-block; width: 10px; height: 10px; margin: 0 6px 0 4px; vertical-align: -1px; border-radius: 2px;
}
.earthcell-content .ec-readout .ec-swatch--forest { background: rgb( 46, 139, 87 ); }
.earthcell-content .ec-readout .ec-swatch--open { background: rgb( 222, 205, 150 ); }
/* The range input covers the whole picture, invisible, so a drag anywhere moves the split. */
.earthcell-content .ec-readout .ec-compare__range {
	position: absolute; inset: 0; width: 100%; height: 100%; margin: 0; opacity: 0; cursor: ew-resize;
	-webkit-appearance: none; appearance: none;
}
.earthcell-content .ec-readout .ec-compare:has( .ec-compare__range:focus-visible ) { outline: 2px solid var(--ec-lime); outline-offset: 3px; }
/* wpautop can wrap the input in a paragraph; the paragraph must not take space. */
.earthcell-content .ec-readout .ec-compare > p { margin: 0; position: static; }
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-compare__caption {
	margin: 12px 0 0; max-width: none; font-family: var(--ec-mono); font-size: 11px; line-height: 1.5;
	letter-spacing: 0.04em; color: var(--ec-quiet);
}

@media ( max-width: 1100px ) {
	.earthcell-content .ec-readout .ec-readings { grid-auto-flow: row; grid-template-columns: repeat( 2, minmax( 0, 1fr ) ); border-bottom: 0; }
	.earthcell-content .ec-readout .ec-reading { padding: 20px 20px 22px 0; border-bottom: 1px solid var(--ec-hair); }
	.earthcell-content .ec-readout .ec-reading + .ec-reading { border-left: 0; }
	.earthcell-content .ec-readout .ec-reading:nth-child( even ) { padding-left: 20px; border-left: 1px solid var(--ec-hair); }
	.earthcell-content .ec-readout .ec-reading:last-child:nth-child( odd ) { grid-column: 1 / -1; }
}
@media ( max-width: 600px ) {
	.earthcell-content > .ec-readout { padding-top: 48px; }
	.earthcell-content .ec-readout .ec-compare { aspect-ratio: 4 / 3; margin-top: 32px; }
	.earthcell-content .ec-readout .ec-reading__value { font-size: 28px; }
	/* One reading per row on a phone: two columns left about 140px of text each. */
	.earthcell-content .ec-readout .ec-readings { grid-template-columns: minmax( 0, 1fr ); }
	.earthcell-content .ec-readout .ec-reading,
	.earthcell-content .ec-readout .ec-reading:nth-child( even ) { padding: 18px 0 20px; border-left: 0; }
	.earthcell-content .ec-readout .ec-reading__source { margin-top: 4px !important; padding-top: 0; }
	/* The two date tags would meet in the middle, so the after tag drops to the bottom corner. */
	.earthcell-content .ec-readout .ec-compare__tag { top: 10px; padding: 5px 8px; font-size: 10px; letter-spacing: 0.1em; }
	.earthcell-content .ec-readout .ec-compare__tag--before { left: 10px; }
	.earthcell-content .ec-readout .ec-compare__tag--after { top: auto; bottom: 10px; right: 10px; }
}

/* --------------------------------------------------------------------------
   AFTER THE HEAD OF SUSTAINABILITY REVIEW, 28 October 2026.

     .ec-readout__meaning   one sentence under the readings saying what they mean
     .ec-brief              "EarthCell in brief": what it is, what it is not, why
                            early funding matters. Three short columns near the top.
     .ec-checklist          "The record" as ticks and dates, recorded and not yet
   -------------------------------------------------------------------------- */
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-readout__meaning {
	max-width: var(--ec-measure, 690px); margin: 28px 0 0; font-size: var(--fs-body); line-height: 1.6;
	color: var(--ec-paper); font-weight: 600;
}
.earthcell-content .ec-brief {
	display: grid; grid-template-columns: repeat( 3, minmax( 0, 1fr ) ); gap: 32px;
	margin: 32px 0 40px; padding: 24px 0; border-top: 1px solid var(--ec-hair, rgba( 62, 68, 84, 0.18 ));
	border-bottom: 1px solid var(--ec-hair, rgba( 62, 68, 84, 0.18 ));
}
.earthcell-content .ec-brief .ec-brief__item > p { margin: 0; max-width: none; font-size: 17px; line-height: 1.55; color: var(--ec-ink); }
.earthcell-content .ec-brief .ec-brief__item > p.ec-brief__label {
	margin: 0 0 10px; font-family: var(--ec-mono); font-size: 11px; font-weight: 600; line-height: 1.3;
	letter-spacing: 0.14em; text-transform: uppercase; color: var(--ec-teal);
}
@media ( max-width: 900px ) {
	.earthcell-content .ec-brief { grid-template-columns: minmax( 0, 1fr ); gap: 20px; }
}
.earthcell-content .ec-checklist { list-style: none; margin: 8px 0 24px; padding: 0; max-width: 760px; }
/* Tick, date and words in one row. The words wrap under themselves, never under the tick. */
.earthcell-content .ec-checklist > li {
	position: relative; margin: 0; padding: 10px 0 10px 128px; min-height: 1.5em;
	border-bottom: 1px solid var(--ec-hair, rgba( 255, 255, 255, 0.18 ));
	font-size: 17px; line-height: 1.5; color: inherit;
}
/* The theme's lime arrow on every list item comes off: a forward arrow beside "Not yet" reads as done. */
/* Out-ranks the theme's ".page .entry-content ul:not(...):not(...):not(...) li::before". */
.page .entry-content.earthcell-content ul.ec-checklist.ec-checklist > li::before {
	position: absolute; left: 0; top: 10px; width: 24px; height: auto; margin: 0; font-weight: 800; line-height: 1.5;
	background: none; background-image: none; border-radius: 0;
}
.page .entry-content.earthcell-content ul.ec-checklist.ec-checklist > li.ec-checklist__done::before { content: "\2713"; color: var(--ec-lime); }
.page .entry-content.earthcell-content ul.ec-checklist.ec-checklist > li.ec-checklist__todo::before { content: "\25CB"; color: var(--ec-quiet, #a9cccf); }
.earthcell-content .ec-checklist > li.ec-checklist__todo { color: var(--ec-quiet, #a9cccf); }
.earthcell-content .ec-checklist .ec-checklist__when {
	position: absolute; left: 30px; top: 14px; font-family: var(--ec-mono); font-size: 11px; font-weight: 600;
	line-height: 1.3; letter-spacing: 0.12em; text-transform: uppercase;
}
@media ( max-width: 600px ) {
	/* On a phone the date sits above the words, both beside the tick. */
	.earthcell-content .ec-checklist > li { padding-left: 30px; padding-top: 30px; }
	.earthcell-content .ec-checklist .ec-checklist__when { top: 10px; }
}

/* --------------------------------------------------------------------------
   LA ESMERALDA, 28 October 2026, after the Head of Sustainability review.

     .ec-ladder       the status ladder: eight steps, the current one lit
     .ec-milestones   the proposed funding steps for buying the land
     .ec-roles        who decides and who does what (the governance panel)
     .ec-claim        the wording a funder could use about its contribution
   -------------------------------------------------------------------------- */
.earthcell-content .ec-ladder { margin: 28px 0 8px; }
.earthcell-content .ec-ladder > p.ec-ladder__label,
.earthcell-content .ec-claim > p.ec-claim__label,
.earthcell-content .ec-roles .ec-roles__item > p.ec-roles__who {
	margin: 0 0 12px; max-width: none; font-family: var(--ec-mono); font-size: 11px; font-weight: 600;
	line-height: 1.3; letter-spacing: 0.14em; text-transform: uppercase; color: var(--ec-teal);
}
.earthcell-content .ec-ladder .ec-ladder__steps {
	list-style: none; margin: 0; padding: 0; display: grid; grid-template-columns: repeat( 8, minmax( 0, 1fr ) ); gap: 6px;
}
.earthcell-content .ec-ladder .ec-ladder__step {
	margin: 0; padding: 12px 10px 14px; display: flex; flex-direction: column; gap: 4px;
	border-top: 4px solid rgba( 62, 68, 84, 0.16 ); color: #5d6672;
}
.earthcell-content .ec-ladder .ec-ladder__step::before { content: none; display: none; }
.earthcell-content .ec-ladder .ec-ladder__step--now { border-top-color: var(--ec-teal); background: var(--ec-tint, #e7eef0); color: var(--ec-ink); }
.earthcell-content .ec-ladder .ec-ladder__n { font-family: var(--ec-mono); font-size: 11px; font-weight: 600; }
.earthcell-content .ec-ladder .ec-ladder__name { font-size: 16px; font-weight: 800; line-height: 1.2; }
.earthcell-content .ec-ladder .ec-ladder__what { font-size: 13px; line-height: 1.35; }
@media ( max-width: 1000px ) {
	.earthcell-content .ec-ladder .ec-ladder__steps { grid-template-columns: repeat( 4, minmax( 0, 1fr ) ); }
}
@media ( max-width: 600px ) {
	.earthcell-content .ec-ladder .ec-ladder__steps { grid-template-columns: repeat( 2, minmax( 0, 1fr ) ); }
}

.earthcell-content .ec-milestones { list-style: none; margin: 8px 0 16px; padding: 0; counter-reset: ec-ms; }
.earthcell-content .ec-milestones > li {
	counter-increment: ec-ms; position: relative; margin: 0; padding: 14px 0 14px 44px;
	display: grid; grid-template-columns: minmax( 0, 1fr ) auto; gap: 2px 16px;
	border-bottom: 1px solid var(--ec-hair, rgba( 62, 68, 84, 0.18 ));
}
.earthcell-content .ec-milestones > li::before {
	content: counter( ec-ms ); position: absolute; left: 0; top: 14px; width: 28px; height: 28px;
	display: grid; place-items: center; border-radius: 50%; background: none; border: 1px solid var(--ec-teal);
	font-family: var(--ec-mono); font-size: 12px; font-weight: 600; color: var(--ec-teal);
}
.earthcell-content .ec-milestones .ec-milestones__name { grid-column: 1; font-size: 17px; font-weight: 800; line-height: 1.35; }
.earthcell-content .ec-milestones .ec-milestones__what { grid-column: 1; font-size: 15px; line-height: 1.45; color: #5d6672; }
.earthcell-content .ec-milestones .ec-milestones__budget {
	grid-column: 2; grid-row: 1 / span 2; align-self: center; font-family: var(--ec-mono); font-size: 11px;
	font-weight: 600; letter-spacing: 0.1em; text-transform: uppercase; color: #5d6672; white-space: nowrap;
}
@media ( max-width: 600px ) {
	.earthcell-content .ec-milestones > li { grid-template-columns: minmax( 0, 1fr ); }
	.earthcell-content .ec-milestones .ec-milestones__budget { grid-column: 1; grid-row: auto; margin-top: 4px; }
}
.earthcell-content .ec-hubband__prose > p.ec-milestones__note { font-size: 15px; color: #5d6672; }

.earthcell-content .ec-roles {
	display: grid; grid-template-columns: repeat( 3, minmax( 0, 1fr ) ); gap: 32px; margin: 32px 0 20px;
	padding-top: 24px; border-top: 1px solid var(--ec-hair, rgba( 62, 68, 84, 0.18 ));
}
.earthcell-content .ec-roles .ec-roles__item > p { margin: 0 0 8px; max-width: none; font-size: 17px; line-height: 1.55; }
.earthcell-content .ec-roles .ec-roles__item > p.ec-roles__who { font-size: 12px; }
.earthcell-content .ec-hubband__inner > p.ec-roles__note { font-size: 15px; color: #5d6672; }
@media ( max-width: 900px ) {
	.earthcell-content .ec-roles { grid-template-columns: minmax( 0, 1fr ); gap: 20px; }
}

.earthcell-content .ec-claim {
	margin: 28px 0 28px; padding: 22px 24px; max-width: 760px; background: rgba( 255, 255, 255, 0.45 );
	border-left: 4px solid #28513f;
}
.earthcell-content .ec-claim > p { margin: 0; max-width: none; color: var(--ec-ink); }
.earthcell-content .ec-claim > p.ec-claim__label { color: #28513f; }
.earthcell-content .ec-claim > p.ec-claim__quote { font-size: 19px; line-height: 1.55; font-weight: 600; }
.earthcell-content .ec-claim > p.ec-claim__note { margin-top: 12px; font-size: 15px; line-height: 1.5; }

/* --------------------------------------------------------------------------
   THE BREADCRUMB ON NIGHT, 28 October 2026. Piers: "White breadcrumbs visible
   below hero image". On a page whose map is followed by the readings band, the
   crumb sat on a pale strip between two dark grounds. It now takes the night
   ground, full bleed, so the map, the crumb and the readings read as one.
   -------------------------------------------------------------------------- */
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb {
	background: #05080b;
	box-shadow: 0 0 0 100vmax #05080b;
	clip-path: inset( 0 -100vmax );
	padding-bottom: 4px;
}
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb,
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb .ec-crumb-sep,
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb [aria-current="page"] {
	color: rgba( 244, 248, 249, 0.62 );
}
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb a {
	color: rgba( 244, 248, 249, 0.86 );
}
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb a:hover,
body .entry-content.earthcell-content:has( > section.ec-readout ) > nav.ec-crumb a:focus-visible {
	color: var(--ec-lime);
}
body .entry-content.earthcell-content:has( > section.ec-readout ) > section.ec-readout { padding-top: 40px; }

/* Four roles sit two by two, so none is left alone on a row (La Esmeralda, 29 October 2026). */
@media ( min-width: 901px ) {
	.earthcell-content .ec-roles:has( > .ec-roles__item:nth-child( 4 ) ) { grid-template-columns: repeat( 2, minmax( 0, 1fr ) ); }
}

/* A round or square mark reads tiny at the 76px cap that suits wide logos, so it may fill the
   plate's full inner height (Love Mother Earth Foundation, 29 October 2026). */
.earthcell-content .ec-partner__logo--square { max-height: 88px; }

/* --------------------------------------------------------------------------
   THE PAGE GRID. 30 October 2026, Piers on Furtado: "use a small, deliberate set of consistent
   content widths and alignment lines".

   Every band sits in the same 1240px column, on twelve columns with a 24px gutter. The widths are:

     wide       the whole column (headings, the readings row, the chooser, the team row)
     reading    690px for paragraphs, from the column's left edge
     split      five columns of heading, seven of prose (.ec-hubband--split, already on the grid)
     side       one third, the last four columns, for anything that sits beside text: the chooser
                panel, the contact card, and the third card of the team row all start on that line

   Measured at 1920 before this block, three things were off it. The contact card started 62px left
   of the third line and the chooser panel 57px right of it, because each one had its own fixed width
   (460px and 21.25rem). The readings row had equal boxes but uneven text, because the first box
   drops its left padding: the gaps between the figures were 273px, then 248px. And the line under
   the H2 and the summary under the readings used the old 680px measure, 10px short of the paragraphs.
   -------------------------------------------------------------------------- */
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-readout__lead,
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-readout__meaning { max-width: 690px; }

@media ( min-width: 1101px ) {
	/* Equal text columns: the gutter sits between the boxes and the dividing line in its middle. */
	.earthcell-content .ec-readout .ec-readings { column-gap: 48px; }
	.earthcell-content .ec-readout .ec-reading,
	.earthcell-content .ec-readout .ec-reading:first-child { padding-left: 0; padding-right: 0; position: relative; }
	.earthcell-content .ec-readout .ec-reading + .ec-reading { border-left: 0; }
	.earthcell-content .ec-readout .ec-reading + .ec-reading::before {
		content: ""; position: absolute; top: 0; bottom: 0; left: -24px; border-left: 1px solid var(--ec-hair);
	}
}

@media ( min-width: 900px ) {
	.earthcell-content .ec-hubband .ec-choose {
		grid-template-columns: minmax( 0, 1fr ) max( 300px, calc( ( 100% - 48px ) / 3 ) );
		column-gap: 24px;
	}
}

@media ( min-width: 761px ) {
	.earthcell-content > #enquiry {
		grid-template-columns: minmax( 0, 1fr ) max( 330px, calc( ( 100% - 48px ) / 3 ) );
		column-gap: 24px;
	}
	.earthcell-content > #enquiry > .ec-enquiry { max-width: none; justify-self: stretch; }
	.earthcell-content > #enquiry > .ec-enquiry-head > h2,
	.earthcell-content > #enquiry.ec-hubband--photo > p { max-width: 690px; }
}

/* --------------------------------------------------------------------------
   THE SITE'S OWN TYPE. 30 October 2026, Piers: "we should be using standard C Level fonts on these
   pages" and "it's just got some weird fonts", pointing at /carbon-credit-retirement/.

   Measured on Furtado at 1920 before this block: about thirty distinct text styles, including a
   monospace (Consolas on Windows) for the reading labels and source links, headings at 39px in
   weight 800, and sizes of 11, 13, 15, 15.2, 16, 17, 20 and 25px. The article page uses five, and
   these pages now use the same five, on the theme's own size tokens so they shrink on a phone:

     heading   Epilogue, --fs-heading, 600      section headings, the big figures
     lead      Epilogue, --fs-lead, 600         card and team headings
     body      Mona Sans, --fs-body, 500/700    paragraphs; bold for the summary line and panel heading
     label     Mona Sans, --fs-label            labels, captions, sources, the chooser panel
     button    Epilogue, 18px, 600              the same as the article page's buttons

   !important because these elements have been styled in several earlier passes, some through an
   id and some through a class repeated four times. This block is meant to be the last word on type.
   -------------------------------------------------------------------------- */
.earthcell-content { --ec-mono: "Mona Sans", sans-serif; }

.earthcell-content h2,
.earthcell-content .ec-facts__value,
.earthcell-content .ec-reading__value {
	font-family: Epilogue, sans-serif !important; font-size: var(--fs-heading) !important;
	font-weight: 600 !important; letter-spacing: -0.01em !important; line-height: 1.2 !important;
}

.earthcell-content h3:not( .ec-choose__heading ),
.earthcell-content .ec-enquiry__heading {
	font-family: Epilogue, sans-serif !important; font-size: var(--fs-lead) !important;
	font-weight: 600 !important; letter-spacing: normal !important; line-height: 1.3 !important;
}

.earthcell-content .ec-eyebrow,
.earthcell-content .ec-hubband__prose > p:first-child,
.earthcell-content > #enquiry > p,
.earthcell-content > #enquiry > p a,
.earthcell-content .ec-reading__note,
.earthcell-content .ec-readout__meaning,
.earthcell-content .ec-choose__heading {
	font-family: "Mona Sans", sans-serif !important; font-size: var(--fs-body) !important;
	letter-spacing: normal !important; text-transform: none !important;
}
.earthcell-content .ec-eyebrow { font-weight: 600 !important; }
.earthcell-content .ec-hubband__prose > p:first-child,
.earthcell-content > #enquiry > p,
.earthcell-content .ec-reading__note { font-weight: 500 !important; }
.earthcell-content .ec-readout__meaning,
.earthcell-content .ec-choose__heading,
.earthcell-content > #enquiry > p a { font-weight: 700 !important; }

.earthcell-content .ec-reading__when,
.earthcell-content .ec-reading__source,
.earthcell-content .ec-reading__source a,
.earthcell-content .ec-figure__caption,
.earthcell-content .ec-photoband__caption,
.earthcell-content .ec-facts__label,
.earthcell-content .ec-choose__panel :is( p, span, a, li ),
.earthcell-content .ec-choose__fallback,
.earthcell-content .ec-enquiry label,
.earthcell-content .ec-enquiry label span,
.earthcell-content .ec-enquiry__context,
.earthcell-content .ec-enquiry__context *,
.earthcell-content > #enquiry small,
.earthcell-content > #enquiry > p:last-of-type {
	font-family: "Mona Sans", sans-serif !important; font-size: var(--fs-label) !important;
	letter-spacing: normal !important; line-height: 1.5 !important;
}
.earthcell-content .ec-reading__when,
.earthcell-content .ec-facts__label { font-weight: 600 !important; letter-spacing: 0.08em !important; text-transform: uppercase; }

.earthcell-content .ec-story__play,
.earthcell-content .ec-enquiry__send {
	font-family: Epilogue, sans-serif !important; font-size: 18px !important;
	font-weight: 600 !important; letter-spacing: normal !important;
}

/* LINKS ON THE DARK BANDS. 30 October 2026, Piers: the reading sources "are barely visible". They
   took the site's teal link colour, rgb(0, 90, 105): 2.5 to 1 on the night band, and 1 to 1 (unreadable)
   on the teal band, where "How EarthCell tracks progress" sat. Lime, as the article page uses for
   links on its dark sections: about 16 to 1 on night and 6.5 to 1 on teal. Buttons carry a class,
   so they keep their own colours. */
.earthcell-content :is( .ec-hubband--night, .ec-hubband--teal ) a:not( [class] ) {
	color: var(--ec-lime, #cef17b) !important;
	text-decoration-color: rgba( 206, 241, 123, 0.5 ) !important;
}
.earthcell-content :is( .ec-hubband--night, .ec-hubband--teal ) a:not( [class] ):hover {
	text-decoration-color: currentColor !important;
}

/* THE FIGURES ON A LIGHT GROUND. 30 October 2026, Piers on the From space section: "that whole block
   should be much lighter where the numbers start". The band stays night down to the lead and the
   story button, then turns white from the top of the readings row to the end of the section, so
   the readings, the summary line and the before-and-after slider (where a page has them) read on
   white with the site's ink and teal links.

   HOW. A pseudo-element on the readings row, behind everything, starting at the row's top edge and
   running past the section's end (5000px: the tallest section, Valezim on a phone, is about 2000px; a
   100000px layer made the browser draw nothing) and past both sides of the window. The section clips it and is its
   own stacking context, so the pseudo paints over the night background and under the content. One
   rule for all four project pages, whatever follows the readings. */
.earthcell-content > .ec-readout { position: relative; z-index: 0; overflow: clip; }
.earthcell-content .ec-readout .ec-readings {
	position: relative; border-top: 0; padding-top: 48px; margin-top: 8px;
}
.earthcell-content .ec-readout .ec-readings::after {
	content: ""; position: absolute; z-index: -1; top: 0; left: -100vmax; right: -100vmax; height: 5000px;
	background: #fff;
}
.earthcell-content .ec-readout .ec-readings,
.earthcell-content .ec-readout .ec-readings ~ * { color: var(--ec-ink); }
.earthcell-content .ec-readout .ec-readings { border-bottom-color: rgba( 62, 68, 84, 0.18 ); }
.earthcell-content .ec-readout .ec-reading__when { color: var(--ec-teal) !important; }
.earthcell-content .ec-readout .ec-reading__value { color: var(--ec-ink) !important; }
.earthcell-content .ec-readout .ec-reading__note { color: var(--ec-ink) !important; }
/* Burn orange darkened for white: #ff9a52 is 2.1 to 1 on white, #b8460e is 5.4 to 1. */
.earthcell-content .ec-readout .ec-reading--alert .ec-reading__value { color: #b8460e !important; }
.earthcell-content .ec-readout .ec-readings ~ p,
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-readout__meaning,
.earthcell-content .ec-readout .ec-hubband__inner > p.ec-compare__caption { color: var(--ec-ink) !important; }
.earthcell-content .ec-readout :is( .ec-readings, .ec-readings ~ * ) a:not( [class] ) {
	color: var(--ec-teal) !important; text-decoration-color: rgba( 0, 90, 105, 0.45 ) !important;
}
@media ( min-width: 1101px ) {
	.earthcell-content .ec-readout .ec-reading + .ec-reading::before { border-left-color: rgba( 62, 68, 84, 0.18 ); }
}
@media ( max-width: 1100px ) {
	.earthcell-content .ec-readout .ec-reading,
	.earthcell-content .ec-readout .ec-reading:nth-child( even ) { border-color: rgba( 62, 68, 84, 0.18 ); }
}
