/* --------------------------------------------------------------------------
   THE EARTHCELL SECTION: THE FIVE EXPLAINING PAGES.

   Added 6 September 2026. Piers: "Each of these pages needs a design pass and a
   tidy-up. The hub, for example, should be the place where the whole idea comes
   to life, and you can link through to other things."

   WHAT THIS FILE IS FOR, AND WHAT IT IS NOT FOR. The hub has its own stylesheet,
   css/calculator-hub.css, and its cards are scoped under .calculator-hub and are
   built on the --hub-* tokens declared there. Those tokens do not exist on
   template-article.php, so none of that CSS can be reached from these five
   pages. This file is the article-side half of the same design language: the
   same teal, the same lime rule, the same monospaced label, the same box.

   WHY IT IS NOT ONE FILE WITH THE HUB. calculator-hub.css is 1,600 lines of a
   single page's layout and every rule in it needs a .calculator-hub ancestor.
   Loading it on an article page would ship all of that to style four boxes.

   IT LOADS ONLY ON PAGES UNDER THE EARTHCELL HUB, so the .ec- names below need
   no page scope of their own. See the enqueue in functions.php, which walks the
   page's ancestors rather than naming ids: the ids differ between the practice
   site and the live one, and a list of them would be wrong on one of them.

   THE MEASURE IS 680px, WHICH IS THE THEME'S, NOT A NEW ONE. Article paragraphs
   are capped at 680px inside a column that is 320px in from the left. Every
   block here matches that width and that left edge, so nothing steps out of the
   text column it belongs to.
   -------------------------------------------------------------------------- */

.page-article .entry-content {

	--ec-ink: #3E4454;
	--ec-paper: #F4F8F9;
	--ec-white: #FFFFFF;
	--ec-teal: #005A69;
	--ec-lime: #CEF17B;
	--ec-rule: rgba( 62, 68, 84, 0.2 );
	/*
	 * THE MUTED TONE IS A COLOUR, NOT AN ALPHA. Changed 6 September 2026.
	 *
	 * It was rgba(62,68,84,0.65), which measures 3.56:1 on the paper ground and 3.67:1 on white.
	 * WCAG AA needs 4.5:1, so the lead paragraph's second half, every card description, the
	 * "Read next" label and every caption on all five explaining pages failed it.
	 *
	 * THIS WAS MISSED ONCE, and the way it was missed is worth writing down. The hub's own token
	 * file fixed the same fault on the hub, and a contrast check reported these pages clean at the
	 * same time. That check read each element's ground off its nearest OPAQUE ancestor and never
	 * composited the alpha, so a 65 per cent ink on paper was measured as solid ink on paper and
	 * passed. A contrast check that does not composite alpha is not a contrast check.
	 *
	 * #5A6270 is 4.9:1 on white and 4.7:1 on paper, and still reads as secondary.
	 */
	--ec-quiet: #5A6270;
	--ec-mono: "IBM Plex Mono", ui-monospace, "SFMono-Regular", Menlo, Consolas, monospace;

	--ec-measure: 680px;
	--ec-s-1: 8px;
	--ec-s-2: 16px;
	--ec-s-3: 24px;
	--ec-s-4: 32px;
	--ec-s-5: 48px;
}

/* --------------------------------------------------------------------------
   THE WAY BACK TO THE HUB.

   These five pages are children of /earthcell/ and the theme prints no visible
   breadcrumb, so a reader who arrives on one from a search result has nothing
   telling them a section exists. This is that link, and it sits above the first
   paragraph rather than below the last, because somebody who is lost is lost at
   the top of the page.
   -------------------------------------------------------------------------- */

.ec-crumb {
	max-width: var(--ec-measure);
	margin: 0 0 var(--ec-s-4);
	font-family: var(--ec-mono);
	font-size: 12px;
	letter-spacing: 0.1em;
	text-transform: uppercase;
}

.ec-crumb a {
	color: var(--ec-teal);
	text-decoration: none;
	border-bottom: 1px solid rgba( 0, 90, 105, 0.35 );
	padding-bottom: 2px;
}

.ec-crumb a:hover,
.ec-crumb a:focus {
	border-bottom-color: var(--ec-teal);
}

/* --------------------------------------------------------------------------
   THE OPENING.

   Every one of these pages opens with two paragraphs that answer the title
   before any heading arrives. They were written to do that job and the template
   was printing them at body size, so they read as the start of the argument
   rather than as the answer. This makes them look like the answer.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-lead {
	max-width: var(--ec-measure);
	margin: 0 0 var(--ec-s-5);
	padding-left: var(--ec-s-3);
	border-left: 3px solid var(--ec-lime);
}

.page-article .entry-content .ec-lead p {
	font-size: clamp( 19px, 1.4vw, 22px );
	line-height: 1.55;
	margin: 0 0 var(--ec-s-2);
}

.page-article .entry-content .ec-lead p:last-child {
	margin-bottom: 0;
	color: var(--ec-quiet);
}

/* --------------------------------------------------------------------------
   THE LIMIT.

   Every page in this section ends its argument with a heading that says the
   same thing: here is what this does not do. That is the section's discipline
   and it was set in unbroken body text, which is where a reader skims. The box
   is teal-ruled rather than red, because it is a statement of scope and not a
   warning: the sentences inside it are ones C Level is content to stand behind.

   IT CARRIES ITS OWN HEADING. The h2 goes inside the box, so the box begins
   where the section begins and a reader scrolling past sees one object rather
   than a heading and then a panel.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-limit {
	max-width: var(--ec-measure);
	margin: var(--ec-s-5) 0;
	padding: var(--ec-s-4);
	background: var(--ec-paper);
	border-top: 3px solid var(--ec-teal);
}

.page-article .entry-content .ec-limit h2 {
	margin-top: 0;
	margin-bottom: var(--ec-s-3);
}

.page-article .entry-content .ec-limit p {
	max-width: none;
}

.page-article .entry-content .ec-limit p:last-child {
	margin-bottom: 0;
}

/* --------------------------------------------------------------------------
   THE ONE-LINE ANSWER UNDER A HEADING.

   Three sections of the tracking page open with a single short sentence that
   answers the heading before the argument starts: "Landscape-scale change
   through time." Set at body size it looked like a paragraph that had lost its
   other half. Set like this it does the job it was written for.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-standfirst {
	max-width: var(--ec-measure);
	margin: 0 0 var(--ec-s-3);
	font-size: clamp( 18px, 1.3vw, 21px );
	line-height: 1.4;
	color: var(--ec-teal);
}

/* --------------------------------------------------------------------------
   POINTS.

   Two or three short things the prose already enumerates: the three principles,
   the three kinds of evidence. They were h3 headings with one sentence under
   each, which is a list pretending to be prose.

   TWO ACROSS, NOT THREE. The column is 680px wide, so three boxes would be
   210px each and the headings would break onto three lines. Two columns of 332
   hold a heading on one line, and the rule below spans the third box across
   both so the bottom right is not left empty.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-points {
	max-width: var(--ec-measure);
	margin: var(--ec-s-4) 0 var(--ec-s-5);
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 280px, 1fr ) );
	gap: var(--ec-s-2);
}

/*
 * THE THIRD BOX SPANS BOTH COLUMNS. Three items in a two-column grid leave the
 * bottom right empty, which reads as a box that failed to load. Spanning it is
 * the difference between a ragged edge and a deliberate one. Two items or four
 * are unaffected: :last-child:nth-child(odd) only matches an odd-numbered last
 * box, which is exactly the case that would leave the hole.
 */
.page-article .entry-content .ec-points > .ec-point:last-child:nth-child( odd ) {
	grid-column: 1 / -1;
}

.page-article .entry-content .ec-point {
	padding: var(--ec-s-3);
	background: var(--ec-white);
	border: 1px solid var(--ec-rule);
	border-top: 3px solid var(--ec-lime);
}

.page-article .entry-content .ec-point h3 {
	margin: 0 0 var(--ec-s-1);
	font-size: 19px;
	line-height: 1.25;
}

.page-article .entry-content .ec-point p {
	margin: 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.55;
	color: var(--ec-quiet);
}

/* --------------------------------------------------------------------------
   A LIST THAT WAS WRITTEN AS A SENTENCE.

   Several of these pages carry a run of six or nine items inside one paragraph:
   what we intend to show for every project, what EarthCell funding can pay for.
   A reader looking for one of those items has to read the whole sentence to
   find out it is not there.

   THE SENTENCE STAYS. The lead-in line above the list is the original sentence
   up to its colon, so the page still reads as prose to somebody reading it
   straight through, and scans to somebody who is not.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-list {
	max-width: var(--ec-measure);
	margin: var(--ec-s-3) 0 var(--ec-s-5);
	padding: var(--ec-s-3) var(--ec-s-4);
	background: var(--ec-paper);
	border-left: 3px solid var(--ec-teal);
}

.page-article .entry-content .ec-list p {
	margin: 0 0 var(--ec-s-2);
	max-width: none;
	font-size: 17px;
	color: var(--ec-quiet);
}

/*
 * THE MARKER IS THE THEME'S, NOT A NEW ONE.
 *
 * This file first drew its own small lime square, and the theme won: style.css
 * sizes `.page .entry-content ul:not(...) li:before` at 24px with a
 * circle-arrow image, and that selector is more specific than anything
 * reasonable here. On a phone the result was a 24px block overlapping the first
 * word of every line.
 *
 * The right answer was not more specificity. The site already has a list style,
 * it is used on every other page, and a section that invented its own would
 * look like a different website. So the rules below only cancel the bottom
 * margin on the last item, so the list does not push the panel open, and leave
 * the marker, the size and the spacing alone.
 */
.page-article .entry-content .ec-list ul {
	margin: 0;
}

.page-article .entry-content .ec-list li:last-child {
	margin-bottom: 0;
}

/* --------------------------------------------------------------------------
   A THING THAT DOES NOT EXIST YET.

   Twelve commercial decisions are open, and two of them leave visible holes on
   the guardians page: the sponsorship offer is being designed, and no guardian
   profile has been cleared for publication. Both were set as ordinary
   paragraphs, which reads as though the thing exists and the page forgot to
   link to it.

   THE DASHED EDGE IS THE POINT. It is the visual language of an empty slot, so
   a reader sees at a glance that nothing is being hidden from them, and so does
   whoever reviews this page next. When Daren settles decisions 7, 8 and 9 the
   box comes off and the offer takes its place. A box that says "not yet" is
   easier to find later than a sentence that says it.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-empty {
	max-width: var(--ec-measure);
	margin: var(--ec-s-3) 0 var(--ec-s-5);
	padding: var(--ec-s-3);
	border: 1px dashed var(--ec-rule);
	background: transparent;
}

.page-article .entry-content .ec-empty__label {
	margin: 0 0 var(--ec-s-2);
	font-family: var(--ec-mono);
	font-size: 12px;
	letter-spacing: 0.1em;
	text-transform: uppercase;
	color: var(--ec-teal);
}

.page-article .entry-content .ec-empty p {
	max-width: none;
	font-size: 17px;
}

.page-article .entry-content .ec-empty p:last-child {
	margin-bottom: 0;
}

/* --------------------------------------------------------------------------
   READ NEXT.

   The reason this pass exists. Five pages sat under one hub with no way from
   any of them to any other, so a section that was written as a whole read as
   five orphans. This is the hub's own .ec-more index, in the article column.

   THE LINK IS IN THE HEADING AND COVERS THE WHOLE BOX, the same arrangement as
   the hub cards and for the same reason: the words a search engine reads are
   the destination's own name, and the reader can still hit anywhere in the box.
   Wrapping the box in an anchor would make its anchor text a whole sentence.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-next {
	max-width: var(--ec-measure);
	margin: var(--ec-s-5) 0 0;
	padding-top: var(--ec-s-4);
	border-top: 1px solid var(--ec-rule);
}

.page-article .entry-content .ec-next__label {
	margin: 0 0 var(--ec-s-3);
	font-family: var(--ec-mono);
	font-size: 12px;
	letter-spacing: 0.1em;
	text-transform: uppercase;
	color: var(--ec-quiet);
}

.page-article .entry-content .ec-next__list {
	display: grid;
	grid-template-columns: repeat( auto-fit, minmax( 280px, 1fr ) );
	gap: var(--ec-s-2);
	margin: 0;
	padding: 0;
	list-style: none;
}

.page-article .entry-content .ec-next__item {
	position: relative;
	padding: var(--ec-s-3);
	background: var(--ec-paper);
	border-top: 3px solid var(--ec-lime);
	transition: background 0.18s ease;
}

.page-article .entry-content .ec-next__item:hover {
	background: var(--ec-white);
	box-shadow: 0 6px 18px rgba( 62, 68, 84, 0.1 );
}

.page-article .entry-content .ec-next__item h3 {
	margin: 0 0 var(--ec-s-1);
	font-size: 19px;
	line-height: 1.25;
}

.page-article .entry-content .ec-next__item h3 a {
	color: inherit;
	text-decoration: none;
}

.page-article .entry-content .ec-next__item h3 a::after {
	content: "";
	position: absolute;
	inset: 0;
}

.page-article .entry-content .ec-next__item:hover h3 a {
	text-decoration: underline;
	text-underline-offset: 3px;
}

.page-article .entry-content .ec-next__item p {
	margin: 0;
	max-width: none;
	font-size: 16px;
	line-height: 1.55;
	color: var(--ec-quiet);
}

/*
 * THE PROJECTS LINE, under the boxes. The five explaining pages exist so the
 * three project pages can be about their projects, so the last thing on every
 * one of them points back at the projects.
 */
.page-article .entry-content .ec-next__projects {
	margin: var(--ec-s-3) 0 0;
	max-width: none;
	font-size: 16px;
	color: var(--ec-quiet);
}

/* --------------------------------------------------------------------------
   ON A PHONE.

   The theme drops the 320px left rail below 1024px and the column becomes the
   full width, so the blocks above need no width change. What they do need is
   less padding: 32px of it inside a 358px box leaves 294px of text.
   -------------------------------------------------------------------------- */

@media ( max-width: 600px ) {

	.page-article .entry-content .ec-limit {
		padding: var(--ec-s-3) var(--ec-s-2);
	}

	.page-article .entry-content .ec-point,
	.page-article .entry-content .ec-next__item {
		padding: var(--ec-s-2);
	}

	.page-article .entry-content .ec-lead {
		padding-left: var(--ec-s-2);
	}

	.page-article .entry-content .ec-list {
		padding: var(--ec-s-2);
	}
}

/* ==========================================================================
   THE UX REVIEW OF 7 SEPTEMBER 2026, CHECKED AND THEN APPLIED.

   Piers sent EarthCell-UX-Review.docx with the "What is an EarthCell?" page.
   Thirteen numbered items in three tiers. Every measurement in it was re-taken
   here against the live staging DOM before anything was changed, because a
   review is a proposal and this file has been wrong before.

   ELEVEN OF THE THIRTEEN CHECKED OUT EXACTLY. The two that did not are recorded
   at the foot of this comment, because not implementing something needs a
   reason on the record as much as implementing it does.

   WHAT WAS CONFIRMED, measured at a 1920px viewport rather than the review's
   1130 and 1512, so a few figures differ in size but none in kind:

     no conversion path   11 links on the page: 2 breadcrumbs, 4 social, 4 read
                          next, 1 inline. Zero buttons. The gap between the
                          social block at y 572 and the read-next cards at
                          y 3868 is 3,296px with nothing to click in it.
     four left edges      652 (15 elements), 677 (8), 684 (12), 711 (2)
     five body sizes      42 / 26 / 22 / 20 / 18 / 17 / 16 / 12, with the list
                          items at 20 LARGER than the body copy at 18
     the 17px orphan      exactly one element on the page: "An EarthCell can
                          include:"
     two measures         body 680px at 18px = 83 characters; card body 282px
                          at 16px = 41. A 42-character spread.
     the dead gutter      padding-left: 320px, .post-side absolute and ending
                          at y 618, page 5,127 tall. 88% of the column is empty.
     heading collapse     h1 48px, h2 42px, a ratio of 1.14
     the "is not" grid    three children, two rendered at 282px of text and one
                          at 630px

   THE REVIEW IS RIGHT ABOUT WHAT MATTERS. Its central point is that the layout
   has drifted out of alignment with itself while the writing stayed disciplined,
   and that is exactly what the numbers show.

   TWO ITEMS WERE NOT IMPLEMENTED, AND HERE IS WHY.

   ITEM 4, "Make the whole Read-next card clickable". It already is, and has
   been since the cards were built. `.ec-next__item h3 a::after` is an absolute
   pseudo-element at inset 0 with the card at position: relative, which is the
   exact stretched-link pattern the review recommends adding. Hit-tested three
   points on a card before believing it: the heading, the description and the
   bottom-right corner all resolve to the same one anchor. The review measured
   the anchor's own box at 217x61 inside a 332x203 card and inferred the rest
   was dead. It is not.

   The review's SECOND point in that item does stand, and is done below: there
   was no sign the card was clickable until a mouse hovered it, and hover does
   not exist on a phone. The cards get the same visible cue the hub cards got
   on the same day.

   ITEM 7, "the grid orphans its third card". The third card is not orphaned,
   it is deliberately spanned across both columns by
   `.ec-point:last-child:nth-child(odd)`, with a comment above it explaining
   that three items in a two-column grid otherwise leave the bottom right empty.
   The review misread the mechanism. It did not misread the RESULT: one of three
   matched items rendering at more than double the measure of the other two is
   a real fault, and the review's own item 9 fixes it properly by making the
   three a full-measure list. That is what is done below, so the span rule and
   the two-column grid both retire.

   ONE ITEM IS DEFERRED AND NOT REFUSED. Item 8, the 320px dead gutter, is the
   biggest single piece of wasted space on the page and the review is right
   about it. Its recommended answer is a sticky contents list, which is a new
   component on a template shared by every article on the site. That is Piers's
   call, not a tidy-up, so it is raised with him rather than built here.

   These rules are scoped to `.page-article .entry-content`, so all five
   EarthCell explaining pages get them together. That is deliberate: the fault
   the review found is a system fault, and fixing it on one page would leave
   four pages drifting away from it.
   ========================================================================== */

/* --------------------------------------------------------------------------
   1. THE MEASURE. 680 -> 620.

   Measured with a Range across the real text rather than by dividing by an
   assumed character width: 680px at 18px runs 83 characters a line. The
   comfortable band is 65 to 75. 620 lands at about 76, which is the top of the
   band rather than the middle, and that is on purpose: taking it to 70 would
   mean 573px, and the h2s and the three-up components below start to look
   cramped before the prose starts to read better.
   -------------------------------------------------------------------------- */

.page-article .entry-content {
	--ec-measure: 620px;
}

/* --------------------------------------------------------------------------
   2. ONE PANEL INSET, SO FOUR LEFT EDGES BECOME TWO.

   The page had body copy at 652, card interiors at 677, the "is not" panel's
   text at 684 and blockquotes at 711. A reader does not see four edges. They
   feel the page wobble.

   The panels keep their inset, because a padded panel is a real boundary and
   the review agrees. What they stop doing is each choosing a different one.
   .ec-limit and .ec-list were at 32px and the cards at 24px, so everything goes
   to 24 and both panel interiors land together.

   THE BLOCKQUOTES LOSE THEIR INDENT AND GAIN A RULE. An indent inside an
   already-indented panel was the fourth edge. A left rule says the same thing
   about a quotation without moving it.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-limit,
.page-article .entry-content .ec-list {
	padding: var(--ec-s-3);
}

.page-article .entry-content .ec-limit blockquote {
	margin-left: 0;
	margin-right: 0;
	padding-left: var(--ec-s-3);
	border-left: 3px solid var(--ec-rule);
}

/* --------------------------------------------------------------------------
   3. THREE BODY SIZES, WITH THE HIERARCHY THE RIGHT WAY UP.

   The list items were 20px, which made a list of package contents the largest
   running text on the page: 11% larger than the argument that frames it. The
   list is reference material and the prose is the persuasion, so the prose
   should not be the quieter of the two. The list goes to body size and the
   marker and the indent do the work of setting it apart.

   The 17px line was one element on the whole page and matched nothing, because
   it was an accident. The card answers were 16px, which was legible but made
   the answers look like footnotes to their own questions.

   Everything a reader reads at length is now 18px. Card metadata stays at 16.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-list li,
.page-article .entry-content .ec-list p,
.page-article .entry-content .ec-point p {
	font-size: 18px;
	line-height: 1.6;
}

/* --------------------------------------------------------------------------
   4. THE "IS NOT" SET BECOMES THREE ROWS, NOT A GRID WITH A SPANNED THIRD.

   This is the review's item 9 and it is the best idea in the document.

   Three plainly-worded "No" answers are a matched set. Rendering two of them at
   282px and the third at 630px says one of them matters more, which is not
   true. Boxing them says they are fine print, which is worse: the whole
   commercial value of this section is that it reads as candour rather than as
   a disclaimer.

   So the boxes come off, the three sit at full measure separated by hairline
   rules, and the questions carry the weight instead of the borders. The 130px
   of empty space that equal-height grid rows left inside the first card goes
   with them.

   THE SPAN RULE IS EXPLICITLY UNDONE rather than deleted from its own block,
   so the comment above it stays with the reasoning that was true while it was
   a grid.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-points {
	display: block;
	margin: var(--ec-s-4) 0 var(--ec-s-5);
	border-top: 1px solid var(--ec-rule);
}

.page-article .entry-content .ec-points > .ec-point:last-child:nth-child( odd ) {
	grid-column: auto;
}

.page-article .entry-content .ec-point {
	padding: var(--ec-s-3) 0;
	background: none;
	border: 0;
	border-bottom: 1px solid var(--ec-rule);
}

.page-article .entry-content .ec-point h3 {
	margin: 0 0 var(--ec-s-1);
	font-size: 21px;
	line-height: 1.3;
}

/* --------------------------------------------------------------------------
   5. THE HEADING SCALE. 48 / 42 was a ratio of 1.14.

   The h1 leaves the viewport about 120px into a 5,127px scroll. From there
   every 42px h2 reads as a page title rather than as a section inside one
   argument, so the page feels like five short pages stacked. The h2s lose
   nothing at 32: they are short and set heavy. What they gain is behaving like
   section markers.
   -------------------------------------------------------------------------- */

.page-article .entry-content h2 {
	font-size: 32px;
	line-height: 1.2;
}

@media ( max-width: 980px ) {

	.page-article .entry-content h2 {
		font-size: 24px;
	}
}

/* --------------------------------------------------------------------------
   6. THE METADATA STOPS SHOUTING IN A WHISPER.

   Published and Last reviewed were 11px in the full heading colour: small
   enough to be hard to read and dark enough to demand attention at the same
   time. This is a hierarchy fix and not an accessibility one; the contrast was
   fine either way.

   THE COLOUR IS WRITTEN OUT RATHER THAN TAKEN FROM --ec-quiet, because
   .post-side sits outside .entry-content and the tokens are declared on
   .entry-content. It is the same #5A6270, and if that value ever moves this
   rule has to move with it.
   -------------------------------------------------------------------------- */

.page-article .post-side__label {
	font-size: 12px;
	color: #5A6270;
}

/* --------------------------------------------------------------------------
   7. THE CARDS SAY THEY ARE CLICKABLE.

   They always were. See the note at the top of this block: the stretched link
   is already there and hit-tests correctly across the whole card. What was
   missing was any sign of it before a hover, which on a phone means no sign at
   all.

   Same cue as the hub cards, for the same reason, on the same day. It is a
   span in the markup rather than an anchor, so the card keeps exactly one link
   and Google keeps one destination name as its anchor text.
   -------------------------------------------------------------------------- */

/*
 * THE CUE CARRIES ITS OWN PARAGRAPH IN THE MARKUP, and that is not decoration.
 * A bare span on its own line is exactly what wpautop wraps, so WordPress would
 * generate the paragraph anyway and this rule would have nothing to target. It
 * is written in the body so the class is there to style. Same lesson as the hub
 * cards, learned the same day.
 */
.page-article .entry-content .ec-next__cue-row {
	margin: 0;
	max-width: none;
}

.page-article .entry-content .ec-next__cue {
	display: inline-flex;
	align-items: center;
	gap: 8px;
	margin-top: var(--ec-s-2);
	padding-bottom: 2px;
	font-size: 15px;
	font-weight: 700;
	line-height: 1.2;
	color: var(--ec-teal);
	border-bottom: 2px solid var(--ec-lime);
}

.page-article .entry-content .ec-next__cue::after {
	content: "\2192";
	font-size: 17px;
	line-height: 1;
	transition: transform 0.15s ease;
}

.page-article .entry-content .ec-next__item:hover .ec-next__cue::after {
	transform: translateX( 3px );
}

/* --------------------------------------------------------------------------
   8. THE CALL TO ACTION THE PAGE DID NOT HAVE.

   The review's highest-value item, and its reasoning is right: a definitional
   page is mid-funnel, and "What can a company say?" is the point where a
   reader stops being curious and starts being qualified. That moment was
   followed by 3,296px of prose and no way to act.

   It is the site's own button shape rather than a new one. There are two on
   the page and no more: one at the point of intent and one under the read-next
   cards, where the trailing sentence used to be the strongest link on the page
   styled as the weakest.

   NO STICKY BAR, which the review also warns against. The page's composure is
   its main asset.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-cta {
	max-width: var(--ec-measure);
	margin: var(--ec-s-4) 0 var(--ec-s-5);
}

.page-article .entry-content .ec-btn {
	display: inline-flex;
	align-items: center;
	gap: 10px;
	padding: 14px var(--ec-s-4);
	font-size: 17px;
	font-weight: 700;
	line-height: 1.2;
	text-decoration: none;
	background: var(--ec-lime);
	color: var(--ec-ink);
	border-bottom: 3px solid var(--ec-ink);
	transition: background 0.15s ease;
}

.page-article .entry-content .ec-btn::after {
	content: "\2192";
	font-size: 18px;
	line-height: 1;
}

.page-article .entry-content .ec-btn:hover,
.page-article .entry-content .ec-btn:focus {
	background: #00EED2;
	color: var(--ec-ink);
}

/* --------------------------------------------------------------------------
   9. GREEN GETS ONE JOB.

   It was doing five: the nav button, the list bullet arrows, the card top
   borders, the lede left rule and the social icon fills. An accent that means
   everything signals nothing, and the page has just gained two real buttons
   that need it to mean something.

   So lime now means "you can click this". The decorative borders take the
   hairline rule, the lede rule takes teal like the other panel rules already
   do, and the list markers stop being arrows.

   THE ARROWS WERE THE WORST OF IT. img/circle-arrow-right.svg is the theme's
   own list marker and it is a navigation affordance. Seven of them on one page,
   none clickable, teaches the eye to discount arrows, which is a tax on the
   two buttons above. A dot says "item in a list" and claims nothing.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-next__item {
	border-top-color: var(--ec-rule);
}

.page-article .entry-content .ec-lead {
	border-left-color: var(--ec-teal);
}

.page-article .entry-content .ec-list li::before {
	width: 18px;
	height: auto;
	margin-top: 0;
	background-image: none;
	content: "\2022";
	font-size: 18px;
	line-height: 1.6;
	color: var(--ec-quiet);
	text-align: left;
}

/*
 * A NUMBERED LIST KEEPS ITS NUMBERS AND LOSES THE DOT. Piers's EarthCell for business rewrite of
 * 11 September 2026 has the first numbered list in this section ("three agreements"). The dot
 * above was drawn after each number, so every item read "1. •the project work". Naming the ol
 * outranks the rule above whatever order the stylesheets load in.
 */
.page-article .entry-content .ec-list ol li::before {
	content: none;
	display: none;
}

/* ==========================================================================
   THREE OF THE RULES ABOVE CHANGED NOTHING, AND THE REASON IS LOAD ORDER.

   Deployed, verified on the page, and three of them had not moved: the h2 was
   still 42px, the list items still 20px, and the body still ran 83 characters
   at a 680px measure although --ec-measure was correctly computing 620.

   THE STYLESHEETS LOAD IN THIS ORDER, read off the rendered page:

     1. style.css
     2. earthcell-section.css      <- this file
     3. earthcell-tokens.css
     4. carbon-calculator.css
     5. single-post.css

   So this file loads SECOND of five. Anything it writes can be overridden by
   three files that come after it, and one of them, single-post.css, is the
   site's blog reading layout, which sets the article measure for every page
   carrying the .single-post body class. This page carries it: the body class
   list is "... no-sidebar single-post page-article earthcell-section ...".

   WHAT WAS BEATING WHAT:

     the h2      .single-post .site-content .site-main article .entry-content h2
                 in style.css, four classes and two elements, against this
                 file's .page-article .entry-content h2 at two and one
     the li      .page .entry-content ul:not(...):not(...):not(...) li in
                 style.css, where each :not() adds its argument's weight
     the measure .single-post .site-content .site-main article .entry-content > p
                 in single-post.css, which both out-specifies this file AND
                 loads after it, so it wins twice over

   THE SELECTORS BELOW USE TWO BODY CLASSES AT ONCE. .earthcell-section and
   .single-post are both genuinely on the body of these five pages, so pairing
   them is a description of the page rather than a trick to inflate a number. It
   takes these rules to five classes, which is one more than the rules they have
   to beat.

   THE REAL FIX IS TO ENQUEUE THIS FILE AFTER single-post.css, which would make
   every rule in it behave the way its author expected. That is one line in
   functions.php, and it would change the outcome of every collision in this
   file at once, including ones nobody has noticed yet. Worth doing deliberately
   and testing across all five pages, rather than as a footnote to a UX review.
   ========================================================================== */

/* --------------------------------------------------------------------------
   THE MEASURE, AND A FINDING BIGGER THAN THIS PAGE.

   single-post.css sets the article measure to 680px, and its own comment gives
   the reasoning: "The text column ran the full 920px at 20px, which is 87
   characters a line. The eye starts losing its place on the return sweep much
   past 75. 680px gives about 71."

   That arithmetic is correct FOR 20px TEXT. The site's body copy is 18px: the
   body element computes to 18px and every paragraph inherits it. At 18px, 680px
   runs 83 characters, measured with a Range across the real text rather than
   assumed. So the column and the type were calibrated against each other at
   different sizes, and the 71 characters the theme intended have been 83 ever
   since.

   THIS IS NOT AN EARTHCELL FAULT. Every article on the site shares the same
   column and the same 18px, so every article runs 83. Fixing it properly is one
   number in single-post.css, or two more pixels on the body, and either is a
   site-wide typographic decision. Raised with Piers on 7 September 2026.

   620px takes these five pages to 76 characters: the top of the comfortable
   band rather than the middle of it. 582px would deliver the theme's own target
   of 71, and is not used, because it would make these five pages visibly
   narrower than every other article on the site while the underlying question
   is still open. Closing most of a gap without inventing a second answer.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .site-content .site-main article .entry-content > p,
.earthcell-section.single-post .site-content .site-main article .entry-content > ul,
.earthcell-section.single-post .site-content .site-main article .entry-content > ol,
.earthcell-section.single-post .site-content .site-main article .entry-content > h2,
.earthcell-section.single-post .site-content .site-main article .entry-content > h3,
.earthcell-section.single-post .site-content .site-main article .entry-content > blockquote {
	max-width: var(--ec-measure);
}

/* The section headings, at the weight needed to beat style.css. */

.earthcell-section.single-post .site-content .site-main article .entry-content h2 {
	font-size: 32px;
	line-height: 1.2;
}

@media ( max-width: 980px ) {

	.earthcell-section.single-post .site-content .site-main article .entry-content h2 {
		font-size: 24px;
	}
}

/* --------------------------------------------------------------------------
   THE LIST ITEMS COME DOWN TO BODY SIZE.

   They were 20px against 18px body copy, which made a list of package contents
   the largest running text on the page. The review's point stands and is worth
   restating precisely: the list is reference material and the prose is the
   persuasion, so the prose should not be the quieter of the two.

   Note that 20px is the theme's list default and 18px is the site's body size,
   which is the same disagreement as the measure, showing up in a second place.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .entry-content .ec-list ul li,
.earthcell-section.single-post .entry-content .ec-list p {
	font-size: 18px;
	line-height: 1.6;
}

/* ==========================================================================
   THE LAST THREE PIXELS, AND THE LIST.

   After the two blocks above, the page measured four left edges rather than the
   two it was meant to have:

     652   body copy, 21 elements
     676   the "is not" panel's interior
     679   the lede and the list interiors
     703   the blockquotes inside the panel

   THE 3px IS THE WHOLE STORY OF 679. .ec-lead and .ec-list carry a 3px left
   rule and 24px of padding, so their text starts 27px inside a panel whose
   siblings start 24px inside. Three pixels is not visible as three pixels. It
   is visible as a column that does not quite line up, which is the fault the
   review named and could not point at.

   So the rule is counted as part of the inset rather than added to it: 3px of
   border plus 21px of padding is the same 24px every other panel uses, and
   every panel interior lands on one line.

   THE BLOCKQUOTES GET A HANGING RULE. The version above gave them a rule and a
   24px padding inside a panel that was already inset, which is how 703 was
   built. They now pull left by one inset and pad back by one inset less the
   border, so the quotation's text sits on the panel's own text edge and the
   rule hangs in the panel's gutter beside it. A quotation should be marked, not
   moved.
   ========================================================================== */

.earthcell-section.single-post .entry-content .ec-lead,
.earthcell-section.single-post .entry-content .ec-list {
	padding-left: calc( var(--ec-s-3) - 3px );
}

.earthcell-section.single-post .entry-content .ec-limit blockquote {
	margin-left: calc( -1 * var(--ec-s-3) );
	margin-right: 0;
	padding-left: calc( var(--ec-s-3) - 3px );
	border-left: 3px solid var(--ec-rule);
}

/* --------------------------------------------------------------------------
   THE LIST ITEMS, AT THE WEIGHT IT ACTUALLY TAKES.

   Second attempt. The rule to beat is in style.css:

     .page .entry-content ul:not(.category-filters):not(.clevel-subnav):not(.flight-details) li

   Each :not() carries its own argument's weight, so that is five classes and
   two elements. The previous attempt had four classes and lost.

   .ec-list is named twice to reach five. That is a tie rather than a win, and a
   tie is enough here: style.css loads first of the five stylesheets and this
   file loads second, so on equal specificity this one is later and takes it.
   Written down because it is the one rule in this block that depends on load
   order rather than surviving it.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .entry-content .ec-list.ec-list ul li {
	font-size: 18px;
	line-height: 1.6;
}

/* ==========================================================================
   THE TITLE AND THE TEXT WERE ON DIFFERENT LEFT EDGES, AND MY LAST AUDIT
   COULD NOT SEE IT.

   Piers pointed at it with two arrows: the breadcrumb, and the gap between the
   H1's left edge and the paragraph under it.

   HOW I MISSED IT. The alignment audit I ran queried
   `.entry-content *` and reported "two left edges, 652 and 676". Both numbers
   were correct and the conclusion was worthless, because the H1 and the
   breadcrumb are not inside .entry-content. They are in .page-article__head,
   which is a sibling. I measured the body of the page and reported on the page.
   An audit that cannot see the largest element on the screen is not an audit.

   WHAT IT ACTUALLY MEASURES, at 1920, taking the ink edge with a Range rather
   than the box:

     632   the breadcrumb and the H1
     652   the body copy
     676   the opening paragraph

   Three edges in the first three elements a reader sees.

   THE CAUSE IS TWO POSITIONING SCHEMES, NOT A STRAY NUMBER. The two blocks are
   siblings inside the same <article> and each centres itself independently:

     .page-article__head   max-width 680,  margin 0 auto, padding 0 20px
                           -> (1905 - 720) / 2 + 20 = 632
     .entry-content        max-width 1240, margin 0 auto, padding-left 320
                           -> (1905 - 1240) / 2 + 320 = 652

   Because one is centred on 720 and the other on 1240, the gap between them is
   a function of the viewport: it is 20px at 1920 and a different number at
   every other width. No single padding value could have fixed it. The head has
   to be given the same box as the body, which is what this does, including at
   the two breakpoints where style.css changes that box: 1290, where the 1240
   cap gives way to a fluid margin, and 980, where the 320px rail closes.

   THE OPENING PARAGRAPH LOSES ITS RULE. .ec-lead carried a 3px left border and
   21px of padding, which put the page's first sentence 24px right of every
   sentence after it. A hanging rule was the other option and is worse here: it
   would sit 24px LEFT of the title, so the alignment problem would move rather
   than go. The paragraph is already set at 22px against an 18px body, which is
   what marks it. The rule was decoration that cost the page its left edge, and
   it is also one of the five jobs the review wants green to stop doing.
   ========================================================================== */

.earthcell-section.page-article .page-article__head {
	max-width: 1240px;
	margin-left: auto;
	margin-right: auto;
	padding-left: 320px;
	padding-right: 20px;
}

/*
 * 1290 is style.css's own breakpoint for the 1240 container: the cap gives way
 * and the block goes fluid with a 15px right margin. The head follows it, or it
 * would keep centring on 1240 while the body stopped.
 */
@media screen and ( max-width: 1290px ) {

	.earthcell-section.page-article .page-article__head {
		max-width: 100%;

		/*
		 * BOTH MARGINS, NOT ONE. The first version of this copied margin-right and left the
		 * left margin on auto, so below 1290 the head centred itself in the remaining space
		 * while the body sat at a flat 15px. Measured at 1280: head 320, body 335. Fifteen
		 * pixels, which is the fault this block exists to remove, reintroduced at one
		 * breakpoint by copying half a rule. Found by testing eight widths instead of one.
		 */
		margin-left: 15px;
		margin-right: 15px;
	}
}

/*
 * 980 is style.css's own breakpoint for the rail: `padding-left: 0` on
 * .entry-content. Same here, or the title would keep a 320px indent on a phone
 * while the text ran full width under it.
 */
@media screen and ( max-width: 980px ) {

	.earthcell-section.page-article .page-article__head {
		padding-left: 0;
	}
}

/* The opening paragraph joins the left edge. */

.earthcell-section.single-post .entry-content .ec-lead {
	padding-left: 0;
	border-left: 0;
}

/* THE BREADCRUMB IS STYLED IN css/earthcell-tokens.css, NOT HERE.

   It was briefly written in both files on 7 September 2026, which is the same fault
   this section keeps finding elsewhere: two places setting one thing, waiting to
   disagree. The tokens file wins the cascade on these pages, so it owns the trail and
   this file says so rather than competing with it. */

/* ==========================================================================
   THE LEFT RAIL GETS A CONTENTS LIST, AND STICKS.

   The review's item 8, and Piers on 7 September 2026: "Yes, we should have a
   sticky contents list."

   WHAT IT WAS. The article template reserves a fixed 320px column and put the
   dates and four share buttons in it. Measured on this page: the block ended at
   y 618 on a page 5,127px tall. So 88 per cent of a column occupying 28 per
   cent of the usable width was empty, and it did not read as restful space
   because it is bounded on both sides. It read as something that failed to load.

   THE LIST WAS ALREADY BEING BUILT. template-article.php has called
   clevel_article_contents() since it was written, to put ids on the H2s, and
   threw away the contents list the same function returns. Nothing new is
   invented here; the second return value is simply kept and rendered.

   STICKY, NOT FIXED. position: sticky keeps the column in view while its parent
   is on screen and lets it go when the article ends, which is what a contents
   list should do. 120px of top offset clears the fixed site header, which is
   the same number css/article-sidebar.css uses for the same reason on the other
   article template.

   IT IS QUIET ON PURPOSE. 14px, muted, no fill, no card, a hairline down the
   left edge. It is a map of the page, not a second navigation, and the eye
   should stay in the text column.

   THE SHARE BLOCK IS LEFT ALONE FOR NOW. The review argues it should drop to
   LinkedIn and Email as text links, on the grounds that share-to-Facebook is
   close to zero value on a B2B explainer and four filled lime circles make it
   the second-loudest thing on the page after the H1. That is a fair argument
   and it changes every blog post on the site, so it is Piers's call and not a
   side effect of adding a contents list.
   ========================================================================== */

/*
 * STICKY ON THE INNER BLOCK, NOT ON THE COLUMN, AND THE FIRST VERSION OF THIS
 * BROKE THE PAGE.
 *
 * .post-side is absolutely positioned by style.css, which is how it sits in the
 * 320px rail without taking any flow space: the rail itself is padding on
 * .entry-content. Setting position: sticky on it put it back into the flow, so
 * it claimed its own height at the top of the article and pushed everything
 * down. Measured: the first paragraph moved from y 266 to y 846. The contents
 * list worked and the page was 580px taller with a hole at the top of it.
 *
 * The column stays absolute and is stretched to the full height of the article
 * with top: 0 and bottom: 0. The block inside it is what sticks, which it can
 * do because it now has a tall parent to travel down. No flow space is taken
 * and nothing moves.
 */
.earthcell-section.single-post .entry-content .post-side {
	position: absolute;
	top: 0;
	bottom: 0;
}

.earthcell-section.single-post .entry-content .post-side__inner {
	position: sticky;
	top: 120px;
}

.earthcell-section.single-post .post-side__contents {
	margin: 0 0 32px;
	padding-left: 16px;
	border-left: 2px solid rgba( 62, 68, 84, 0.16 );
}

.earthcell-section.single-post .post-side__contents .post-side__label {
	display: block;
	margin: 0 0 12px;
	font-family: Epilogue, sans-serif;
	font-size: 12px;
	font-weight: 700;
	letter-spacing: 0.06em;
	text-transform: uppercase;
	color: #5A6270;
}

.earthcell-section.single-post .post-side__toc {
	margin: 0;
	padding: 0;
	list-style: none;
}

.earthcell-section.single-post .post-side__toc li {
	margin: 0 0 10px;
	padding: 0;
	line-height: 1.35;
}

.earthcell-section.single-post .post-side__toc li::before {
	content: none;
}

/*
 * THE THEME MAKES LIST LINKS BOLD AT A 30px LINE, which is right for a
 * paragraph and wrong for a stack of eight short entries. Named at four classes
 * because the rule it has to beat carries three :not() selectors and their
 * weight, the same collision the body list ran into higher up this file.
 */
.earthcell-section.single-post .entry-content .post-side__toc li a {
	display: block;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	font-weight: 400;
	color: #5A6270;
	text-decoration: none;
	border: 0;
}

.earthcell-section.single-post .entry-content .post-side__toc li a:hover,
.earthcell-section.single-post .entry-content .post-side__toc li a:focus {
	color: var(--ec-ink, #3E4454);
	text-decoration: underline;
	text-underline-offset: 3px;
}

/*
 * THE HEADINGS NEED SOMEWHERE TO LAND. Without this an anchor puts the heading
 * under the fixed site header, which is the fault the hero clearance rule in
 * earthcell.css already records for the H1. Same number.
 */
.earthcell-section.single-post .entry-content h2[id] {
	scroll-margin-top: 110px;
}

/*
 * BELOW 980 THE RAIL CLOSES, which is style.css's own breakpoint, so the column
 * stops being a column and the list would sit above the article as a stack of
 * links nobody asked for. It comes off, and the dates and share block behave as
 * they did before.
 */
@media screen and ( max-width: 980px ) {

	.earthcell-section.single-post .post-side__contents {
		display: none;
	}

	.earthcell-section.single-post .entry-content .post-side {
		position: static;
		bottom: auto;
	}

	.earthcell-section.single-post .entry-content .post-side__inner {
		position: static;
	}
}

/* --------------------------------------------------------------------------
   TWO THINGS I COULD SEE IN THE SCREENSHOT AND HAD NOT MEASURED.

   THE CONTENTS LIST HAD THE THEME'S GREEN ARROWS. img/circle-arrow-right.svg
   is applied by the theme to li::before across the site, so the four entries
   came out as 24px lime circles beside 14px text: the loudest thing in a column
   that is meant to be quiet, and the same arrow the review wants off the body
   list because an arrow is a navigation affordance. Here the links really are
   navigation, which makes it worse rather than better: four arrows pointing at
   a list of four arrows.

   `content: none` at three classes lost to the theme's rule. Four classes and
   the width zeroed as well, because the pseudo-element carries a fixed 24px
   box that would otherwise keep its indent after the picture had gone.

   THE LEDE HAD 96px UNDER IT, NOT THE 48 ITS OWN RULE ASKS FOR. Something
   doubles it further down the cascade. It was tolerable while the lede had a
   left rule holding it together as a block; with the rule gone it read as a
   hole between the opening and the argument. Measured: lead bottom 496, next
   paragraph top 592.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .entry-content .post-side__toc li::before {
	content: none;
	display: none;
	width: 0;
	height: 0;
	background-image: none;
}

/* The lede's bottom gap is set in css/earthcell-tokens.css, which owns this
   section's spacing tokens and loads after this file. Trying to set it here was
   the same two-places-for-one-thing fault this file keeps finding: the rule was
   valid, matched, and lost. */

/* ==========================================================================
   THE BREADCRUMB WAS UNDER THE NAVIGATION BAR. THAT IS WHAT "SQUASHED" MEANT.

   Piers said it first and I fixed the wrong thing. I changed the trail's
   typeface and case, which needed doing, and never looked at where it sat.

   MEASURED AT 1920, WHICH IS THE WHOLE FAULT IN TWO NUMBERS:

     the fixed header's bottom edge   y 73
     the breadcrumb's top edge        y 70

   Minus three pixels of clearance. The first line of the page began three
   pixels underneath the navigation, so it read as crushed against it, and the
   arrow in the screenshot pointed exactly there.

   WHY THE CLEARANCE RULE THAT EXISTS DID NOT APPLY. css/earthcell.css carries
   `.earthcell-section.page-article .page-article__head { padding-top: 96px }`
   with a long note about the fixed header sitting over the page rather than
   above it. That file is the EarthCell TEMPLATE stylesheet and is not loaded on
   article pages: these five load style.css, earthcell-section.css,
   earthcell-tokens.css, carbon-calculator.css and single-post.css. So the fix
   was written, and written up, in a file this page never sees.

   What the page actually gets is `margin: 70px auto 0` from single-post.css,
   which was set when the article opened with an H1 and no trail above it. 70px
   of margin against a 73px header leaves nothing.

   THE NUMBER IS 48px OF PADDING ON TOP OF THAT 70. It gives 45px of clear air
   between the navigation and the trail, which is the same order as the gap
   between the trail and the H1, so the two read as one block introducing the
   page rather than as something jammed into a corner.
   ========================================================================== */

.earthcell-section.page-article .page-article__head {
	padding-top: 48px;
}

/*
 * BELOW 980 single-post.css drops the head's margin to 40px, so the padding has
 * to do more of the work rather than less.
 *
 * 32px was the first guess and it was wrong: measured at 390, the header's
 * bottom edge is 68 and 40 + 32 put the trail's top at 72. Four pixels, which
 * is the same fault as the desktop one at a smaller scale, on the width most
 * readers use. Guessing a mobile number from a desktop number is what produced
 * it; 56 was measured, and gives 28px of clear air at 390.
 */
@media screen and ( max-width: 980px ) {

	.earthcell-section.page-article .page-article__head {
		padding-top: 56px;
	}
}

/* --------------------------------------------------------------------------
   THE CONTENTS LIST MOVED BELOW THE DATES, SO ITS RULE MOVES TO THE TOP.

   It had a hairline down its left edge when it opened the column. Sitting
   between the dates and the share block it needs the same separator the date
   blocks use between themselves, which is a rule across the top, or it reads as
   part of the "Last reviewed" block above it.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .post-side__contents {
	margin: 0 0 24px;
	padding: 18px 0 0;
	border-left: 0;
	border-top: 1px solid rgba( 62, 68, 84, 0.15 );
}

/* ==========================================================================
   THE BLACK LINE ABOVE "WHAT CAN A COMPANY SAY?" WAS A BORDER AROUND NOTHING.

   Piers: "the unusual black line above what can a company say?"

   MEASURED, AND THE ANSWER IS IN TWO COLOURS THAT ARE THE SAME:

     the page background    rgb(244, 248, 249)
     the panel background   rgb(244, 248, 249)

   .ec-limit was built as a panel: a paper ground with a 3px teal rule across
   the top. But the page's own ground is that same paper, so the panel has never
   been visible on these pages. The only thing marking it was the rule, and a
   3px bar of #005A69 with nothing under it does not read as the top edge of a
   box. It reads as a line somebody left behind. Teal that dark is close enough
   to black at 3px that Piers called it black.

   IT ALSO STOPPED AGREEING WITH ITS NEIGHBOURS TODAY. The three "is not"
   answers directly above it became a list separated by 1px hairlines at
   rgba(62,68,84,.2). So the heaviest rule on the page now sits immediately
   under the lightest ones, marking a box that is not there.

   SO THE PANEL GOES AND THE SECTION STAYS. It is a section like the one above
   it and it is marked the same way: one hairline, at the same weight and
   colour, with the heading under it. The side padding goes too, which brings
   the section's text back to the page's own left edge at 652 rather than the
   panel interior at 676. One fewer edge on a page whose left edge was the
   thing Piers pointed at this morning.

   NOTHING IS LOST FROM THE CLAIMS BOUNDARY. What made that section careful was
   always the writing. The box was decoration, and an invisible box at that.

   THE EXAMPLE CLAIMS KEEP THEIR RULE AND TAKE A SMALL INDENT. With the panel
   padding gone there is no gutter for a hanging rule to hang in, and a rule
   pulled left of the body edge would be the misalignment this page has just
   spent the day removing. An indented quotation with a rule beside it is a
   form every reader knows, and it is inside a section whose other text sits on
   the page's edge, so it reads as a quotation rather than as drift.
   ========================================================================== */

.earthcell-section.single-post .entry-content .ec-limit {
	background: none;
	padding: 24px 0 0;
	border-top: 1px solid rgba( 62, 68, 84, 0.2 );
}

.earthcell-section.single-post .entry-content .ec-limit blockquote {
	margin-left: 0;
	margin-right: 0;
	padding-left: 20px;
	border-left: 3px solid rgba( 62, 68, 84, 0.2 );
}

/*
 * THE TWO EXAMPLE CLAIMS HAD NO SPACE AROUND THEM, AND TAKING THE PANEL AWAY
 * MADE IT VISIBLE.
 *
 * Measured after the panel came off: both blockquotes compute margin-top 0 and
 * margin-bottom 0, so the first ends at y 365 and the second begins at y 365,
 * and the paragraph after begins at 423 where the second ends at 423. Three
 * elements touching. They read as one run-on block rather than as two separate
 * sentences a company could say.
 *
 * It was wrong before this morning too. The panel's ground and padding gave the
 * whole group an outline, which hid the fact that nothing inside it was spaced.
 * Removing a box does not create this kind of fault; it stops covering it.
 *
 * 20px between them and 32 after the pair, which is the paragraph rhythm this
 * section already uses and a little more air where the examples end and the
 * argument resumes.
 */

.earthcell-section.single-post .entry-content .ec-limit blockquote {
	margin-top: 0;
	margin-bottom: 20px;
}

.earthcell-section.single-post .entry-content .ec-limit blockquote + p {
	margin-top: 12px;
}

/* --------------------------------------------------------------------------
   THE ONE DIAGRAM ON THE SECTION.

   Piers asked for illustrations. The UX review asked for exactly one, and named
   which one: the project-versus-cell distinction, because it is a spatial idea
   carried by four paragraphs of prose.

   THE PICTURE SCALES AND THE WORDS DO NOT. Everything a reader has to read is
   in the figcaption, which is ordinary HTML at 14px. The first version put the
   labels inside the svg and they came out at about seven pixels on a phone.
   A diagram that has to be zoomed is not a diagram.

   IT KEEPS THE PAGE'S MEASURE rather than running full width. The review is
   right that this page's restraint is its main asset, so the picture behaves
   like a paragraph: same left edge, same right edge, no breaking out.
   -------------------------------------------------------------------------- */

.earthcell-section.single-post .entry-content .ec-diagram {
	max-width: var(--ec-measure);
	margin: 40px 0;
	padding: 0;
}

.earthcell-section.single-post .entry-content .ec-diagram svg {
	display: block;
	width: 100%;
	height: auto;
}

.earthcell-section.single-post .entry-content .ec-diagram figcaption {
	max-width: none;
	margin: 14px 0 0;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.5;
	color: #5A6270;
}


/* ==========================================================================
   PICTURES ON THE FIVE EXPLAINING PAGES.
   Added 7 September 2026.

   Piers: "All of the pages, even the informational ones, need images, diagrams,
   maps, or both." Before today the five explaining pages carried none. Measured
   on staging: nought images in the body of every one of them. The only pictures
   in the section were on the hub and the three project pages.

   THE ONE DIAGRAM WAS ALREADY HALF BUILT. `.ec-diagram` was styled in this file
   on 7 September and no page ever carried the markup, so the rule had nothing to
   style. That is fixed here and the diagram is on the page it was drawn for.

   EVERY PHOTOGRAPH IS C LEVEL'S OWN, FROM KOGI OR BIOESTRELA, AND THAT IS A
   RULE AND NOT A PREFERENCE. The media library also holds Khasi Hills, Mikoko
   Pamoja, CommuniTree, Tahiry Honko and the other certified carbon projects. An
   EarthCell page exists to say it is NOT one of those. A nursery in Mozambique
   under a heading about what EarthCell funding pays for would undo, in one
   picture, the boundary the whole section draws in words.

   THE PICTURE KEEPS THE PAGE'S MEASURE. These pages are a reading column with a
   sidebar, not a band system, and the review's finding was that the restraint is
   the page's main asset. So a figure behaves like a paragraph: same left edge,
   same right edge, no breaking out to full width.
   ========================================================================== */

.page-article .entry-content .ec-figure {
	max-width: var(--ec-measure);
	margin: 40px 0;
	padding: 0;
}

.page-article .entry-content .ec-figure img {
	display: block;
	width: 100%;
	height: auto;
	border-radius: 10px;
}

/*
 * THE CAPTION IS THE SAME 14px AS THE DIAGRAM'S, in the same muted tone, because
 * they do the same job and a reader should not have to work out that they are
 * different kinds of note.
 */
.page-article .entry-content .ec-figure figcaption {
	max-width: none;
	margin: 14px 0 0;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.5;
	color: #5A6270;
}

/* --------------------------------------------------------------------------
   THE DIAGRAMS.

   THE PICTURE SCALES AND THE WORDS DO NOT. Everything a reader has to read sits
   in the figcaption as ordinary HTML at 14px. The first version of the one
   diagram put its labels inside the svg, and because the svg scales to the
   column they came out at about seven pixels on a phone. A diagram that has to
   be zoomed is not a diagram.

   SO THE SVGs BELOW CARRY SHAPES AND NUMBERS ONLY. Two-character labels survive
   scaling; sentences do not.

   THE COLOURS ARE THE SECTION'S OWN and are named in the svg rather than here,
   because a shape's fill belongs with the shape. Teal is the thing being
   described, lime is the part under discussion, and the pale slate is context.
   -------------------------------------------------------------------------- */

.page-article .entry-content .ec-diagram {
	max-width: var(--ec-measure);
	margin: 40px 0;
	padding: 0;
}

.page-article .entry-content .ec-diagram svg {
	display: block;
	width: 100%;
	height: auto;
}

.page-article .entry-content .ec-diagram figcaption {
	max-width: none;
	margin: 14px 0 0;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.5;
	color: #5A6270;
}

/*
 * THE KEY UNDER A DIAGRAM. Three items, each a swatch and a phrase, as ordinary
 * text so it stays legible at any width. It wraps rather than scrolls.
 *
 * THE CLASS IS NAMED FOUR TIMES AND HERE IS THE ARITHMETIC.
 *
 * The theme dresses every list in the article body with a green circled arrow:
 * `.page .entry-content ul:not(.category-filters):not(.clevel-subnav):not(.flight-details) li`.
 * The three `:not()` arguments each contribute a class, so that rule is five
 * classes and two elements, and it sets a 24px arrow image, 20px type and a
 * 12px gap.
 *
 * The first version here was three classes and one element and lost every one
 * of those. Measured on staging: the key rendered with the theme's arrows
 * instead of the colour swatches it exists to show, at 20px instead of 14, so a
 * key to a diagram pointed at nothing. Two of the three items looked like links.
 *
 * Four names is six classes and two elements. It wins outright, and the arrow
 * has to be turned off explicitly rather than merely overridden, because a
 * background image is not undone by setting a background colour.
 */
.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key {
	display: flex;
	flex-wrap: wrap;
	gap: 8px 24px;
	margin: 14px 0 0;
	padding: 0;
	list-style: none;
}

.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key > li {
	display: flex;
	align-items: center;
	gap: 8px;
	margin: 0;
	padding: 0;
	font-family: Epilogue, sans-serif;
	font-size: 14px;
	line-height: 1.4;
	color: #5A6270;
}

.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key > li::before {
	content: "";
	flex: 0 0 auto;
	width: 14px;
	height: 14px;
	margin: 0;
	border-radius: 3px;
	background-image: none;
	background-color: var(--ec-swatch, #005A69);
}

.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key > li.is-cell { --ec-swatch: #CEF17B; }
.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key > li.is-project { --ec-swatch: #005A69; }
.page-article .entry-content ul.ec-key.ec-key.ec-key.ec-key > li.is-rest { --ec-swatch: #D5DEE1; }


/* --------------------------------------------------------------------------
   TWO THINGS THAT SHOULD HAVE AGREED AND DID NOT.

   1. THE FIGURES WERE 60px WIDER THAN THE PROSE. Measured at 1440px: every
      paragraph ended at x=1032 and every picture at x=1092. Both start on the
      same left edge, so the eye reads it as a picture that is slightly out
      rather than as a deliberate wider column.

      THE CAUSE IS NOT THE VALUE, IT IS THE SELECTOR. This file already asks for
      `max-width: var(--ec-measure)`, and that token resolves to 620px at the
      element. The theme's own article rule ends
      `.entry-content > figure:not(.alignwide):not(.alignfull)` and sets a flat
      680px from five classes deep, so it beat the three-class rule here and the
      token was never used. Matching that shape is what makes the value apply.

      `--ec-measure` IS DECLARED TWICE ON THE SAME SELECTOR IN THIS FILE, 680px
      and then 620px. The second wins and 620 is the live measure. The first is
      dead and should come out when somebody is next in that block; it is left
      alone today because removing a token this many rules read is its own job.

   2. A PORTRAIT PHOTOGRAPH RAN 942px TALL. Two of them on the Nature Guardians
      page, on a page of 581 words. At 620px wide a 3:4 picture is most of a
      laptop screen, and the reader has to scroll past it to get back to the
      argument. Capping the height rather than the width keeps the shape and the
      left edge, and leaves every wide photograph exactly as it was: none of
      them is tall enough for the cap to touch it.
   -------------------------------------------------------------------------- */

/*
 * THE CLASS IS NAMED TWICE AND THE ELEMENT IS NAMED TOO, AND BOTH ARE NEEDED.
 *
 * The first version of this rule matched the theme's chain and still lost. The
 * theme's selector ends `figure:not(.alignwide):not(.alignfull)`, and a `:not()`
 * contributes the specificity of its argument, so that rule is six classes and
 * two elements. Naming the chain alone is six classes and one element, which is
 * lower, and the picture stayed 60px wider than the prose after a deploy that
 * looked like it had worked.
 *
 * TWO WAS A TIE AND A TIE IS NOT A WIN. Naming the class twice made it six
 * classes and two elements, exactly level with the theme, and style.css loads
 * after this file, so the theme still won and a second deploy changed nothing on
 * the page. The `.ec-key` moved and the figures did not, which is what showed it.
 *
 * Three names is seven classes and two elements. That wins outright and does not
 * depend on which of the five stylesheets the browser reads last.
 */
.single-post .site-content .site-main article .entry-content > figure.ec-figure.ec-figure.ec-figure,
.single-post .site-content .site-main article .entry-content > figure.ec-diagram.ec-diagram.ec-diagram,
.single-post .site-content .site-main article .entry-content > ul.ec-key.ec-key.ec-key {
	max-width: var(--ec-measure);
}

.page-article .entry-content .ec-figure img {
	width: auto;
	max-width: 100%;
	max-height: 620px;
}

/* On a phone the column is already the constraint, so the cap only gets in the
   way: it would leave a tall picture narrower than the words beside it. */
@media screen and ( max-width: 730px ) {
	.page-article .entry-content .ec-figure img {
		width: 100%;
		max-height: none;
	}
}
