/**
 * core/group rules, each scoped to the unit that owns it.
 *
 * This file is shared by construction: every unit that places a core/group lands here, and
 * that is nearly every unit. Nothing in it may be written unscoped -- a rule with no unit
 * hook class repaints every group on the site, which is a theme decision and belongs to
 * wp-theme-scaffold.
 */

/*
 * THE box-sizing RESET THAT USED TO BE HERE IS NOW THEME-WIDE. Do not re-add a per-unit copy.
 *
 * Until 2026-09-11 this file opened with one `box-sizing: border-box` declaration guarded by a
 * sixteen-selector list -- four units x four forms each (`.wm-X`, `.wm-X *`, `.wm-X *::before`,
 * `.wm-X *::after`). It is now one rule in `theme.json`'s `styles.css`, keyed to
 * `.wp-site-blocks`, and every unit gets it without naming itself.
 *
 * WHY IT IS A THEME DECISION AND NOT FOUR UNIT DECISIONS: the source is built on Zurb
 * Foundation 6, which makes every box on the page a border-box box, so EVERY measured box in
 * EVERY contract on this build is a border-box measurement. A theme without the reset disagrees
 * with the reference by construction rather than by accident. Four units reached that conclusion
 * separately, by three distinct mechanisms, and all three are recorded because they are the
 * evidence the theme rule is load-bearing rather than cosmetic:
 *
 *   - `section-content` -- a `max-width` on a heading. Content-box added the root group's 15px
 *     gutters on top of `layout.contentSize`, so the band heading measured 1200 at x=120 where
 *     the contract asserts 1170 at x=135.
 *   - `link-description-cards` and `image-link-cards` -- `flex-basis` on `.wp-block-column`, the
 *     one box core sets border-box on the PARENT (`.wp-block-columns`) and never on the child.
 *     Measured: under content-box a `flex-basis: 33.3333%` over a 1170 content box is 390 PLUS
 *     30px of gutter, so the column's outer box is 420 and only two fit a 1170 line -- a 3-up
 *     grid renders 2-up at 1440 where its contract measures 3-up, with nothing reporting it.
 *   - `listing-hero` -- a percentage `max-width` on a group that places no `core/columns` at all.
 *     Content-box sized the CONTENT box, so the column measured 830 where its contract measures
 *     800 and its heading landed at 800/x=320 instead of 770/x=335.
 *
 * Each of those had an accidental pass at one viewport and a failure at another, which is why
 * none of them is safe to leave to a per-unit rule somebody might not add.
 *
 * WHAT THE PROMOTION WAS MEASURED TO COST, on the four instance pages at all four captured
 * viewports (pages 9, 33, 34 and 40 under the sandbox parent): of 140 nodes inside <main>,
 * exactly ONE per page -- `div.entry-content` -- changed its computed `box-sizing`, and its box
 * did not move. Nothing else in any unit moved by a pixel. The measurement excluded the header
 * and footer `core/navigation`, whose `core/page-list` item count changes between runs while
 * sandbox pages are being created, and checked those separately.
 */

/*
 * .wm-raw-html: the band's 16px of content height is a contained bottom margin on the raw
 * `<p>`, and nothing in block markup can write it.
 *
 * The contract measures this band's root at 1440x76 at all four viewports and resolves it as
 * 30 + 16 + 30, while every descendant -- the `<p>`, the `field__item` and the field wrapper
 * -- measures 1170x**0**, because the field's whole value is an empty anchor target. A 16px
 * parent over a 0px child is a bottom margin the parent contains, and in the source it is
 * Foundation's global `p { margin-bottom: 1rem }` held in by `.row`'s table clearfix. Here
 * the root group's own 30px bottom padding does the containing.
 *
 * It cannot be an attribute. core/html renders its `content` verbatim with no wrapper of its
 * own and declares no spacing support (`supports.className` is `false` and there is no
 * `spacing` key in its block.json), so the `<p>` is a raw element with no block to hang a
 * style on. Writing the margin inline in the block's content instead would work and was
 * rejected: content is what the import replaces, and the margin is the band's rhythm rather
 * than this instance's value.
 *
 * SPECIFICITY, measured rather than assumed. `blockGap: "0"` on the root compiles to
 * `.wp-container-core-group-is-layout-<hash> > * { margin-block-start: 0; margin-block-end: 0 }`
 * -- read off wp_style_engine_get_stylesheet_from_context() on this install, 2026-09-11 -- at
 * 0,1,0. This selector is 0,1,1 and wins on its own. The `:last-child` companion is there
 * because the constrained-layout gap rule grows a `> :last-child` form at 0,2,0 when blockGap
 * is non-zero, and a later instance of this band may not carry `blockGap: "0"`.
 *
 * Reproduced for `p` only. The source has the whole of Foundation's typography applying to
 * whatever markup the field holds; this band has one measured instance, whose markup is a
 * paragraph. A unit whose only capture paints nothing is a thin basis for more than that.
 */
.wm-raw-html > p,
.wm-raw-html > p:last-child {
	margin-block-end: var(--wp--preset--spacing--25);
}

/*
 * .wm-raw-html .ghost-btn / .button-row -- moonshot.yml instance 1, the jump-link bar
 * (sites/washington-mills/ai/import/import-pages.php's wm_build_moonshot_button_row()).
 *
 * NOT A MEASUREMENT. ai/pages/moonshot.yml's own `why` on this instance describes the
 * source's rule in prose only -- "flex row, translucent fill, 1px border, uppercase, a
 * hover that changes the font size as well as the colour" -- it does not quote the
 * capture's literal `<style>` text, and that capture is not something a content-import
 * dispatch may open (CLAUDE.md, the artifact rule). So this is a faithful-to-description
 * DECISION, built from theme tokens rather than invented pixel values, not a reproduction
 * of the source's own numbers. Revisit if the real rule is ever recovered.
 */
.wm-raw-html .button-row {
	display: flex;
	flex-wrap: wrap;
	gap: var(--wp--preset--spacing--20, 10px);
}

.wm-raw-html .ghost-btn {
	display: inline-block;
	padding: var(--wp--preset--spacing--20) var(--wp--preset--spacing--30);
	background-color: rgba(0, 0, 0, 0.05);
	border: 1px solid currentColor;
	text-transform: uppercase;
	text-decoration: none;
	color: inherit;
	transition: font-size 0.2s ease, color 0.2s ease;
}

.wm-raw-html .ghost-btn:hover {
	font-size: 1.1em;
	color: var(--wp--preset--color--primary);
}

/*
 * .wm-raw-html__duraluna .float-img -- moonshot.yml instance 3, the DURALUNA product band
 * (wm_build_moonshot_duraluna_band()). Same caveat as .ghost-btn above: the source's own
 * `why` records only that "the image floats right with the text wrapping around it", not a
 * quoted rule, so the exact margin below is a decision, not a measured figure. The image's
 * own 280x280 is the instance's recorded `image_width`/`image_height`, not a value chosen
 * here.
 */
.wm-raw-html__duraluna .float-img {
	float: right;
	max-width: 280px;
	height: auto;
	margin: 0 0 1.5em 1.5em;
}

@media (max-width: 639px) {
	.wm-raw-html__duraluna .float-img {
		float: none;
		display: block;
		margin: 0 auto 1.5em;
	}
}

/*
 * .wm-news-article: THE ARTICLE COLUMN'S OWN WIDTH, at all four captured viewports, because no
 * block attribute on this theme can express it.
 *
 * WHAT THE CONTRACT MEASURES. The root node carries the box at every viewport in the sweep:
 * "1440x900: 970x318 at x=235; 1024x768: 994x308; 768x1024: 738x308; 390x844: 370x366". Read as
 * gutters against the viewport those are 235+235 (a capped column, centred), 15+15, 15+15 and
 * 10+10 -- a FLUID column with a 15px gutter that stops growing at 970, and a 10px gutter on the
 * phone. The same shape the rest of this build records for the source's Zurb Foundation 6 grid,
 * and the same 10px `@mobile` gutter `.wm-section-content` writes as a block attribute.
 *
 * WHY NOT A BLOCK ATTRIBUTE, which is where a width belongs and where this one cannot go. A
 * constrained layout's `contentSize` caps the CHILDREN of the block that declares it, never the
 * block itself, so the article's own 970 has to come from its parent; and the cap is not in force
 * at 1024, where the source's article is 994 -- WIDER than 970. A single `contentSize` on
 * templates/single.html's `main` would therefore be right at 1440 and 24px too narrow at 1024. The
 * edge the source switches on is Foundation's `xlarge`, 1200px (the same breakpoint
 * assets/css/blocks/button.css records for this source's chrome), and this theme's two viewport
 * bands are 639.98px and 1123.98px -- so `style["@tablet"]` cannot spell it either. A scoped
 * stylesheet writes its own media query, which is the one spelling that is right at all four.
 *
 * 970 IS A MAX-WIDTH AND NOT A WIDTH, and the 1200px query is what the measurements say rather
 * than what Foundation's defaults say: 994 at 1024 puts the cap out of force there, 970 at 1440
 * puts it in force there, and 1200 is the documented Foundation edge between them. Nothing in the
 * projection reads the source's stylesheet for this block, so the EDGE is an inference over two
 * measured widths while both widths are transcriptions. Named in the build hand-back; if the real
 * edge is 1024, the band 1024-1199 renders a fluid column where the source has a capped one.
 *
 * THE THEME'S 40px ROOT GUTTER IS NOT IN THE SOURCE AND IS NOT FOUGHT HERE, it is avoided.
 * `settings.useRootPaddingAwareAlignments` is true and layout.php:1110-1117 puts
 * `has-global-padding` on every constrained container, so the column's `main` in single.html is a
 * FLOW group: a flow group takes no global padding, the article is its own centred column by the
 * rule below, and there is no inherited 40px to subtract. Measured on this install before the
 * change: with a constrained `main` the article came back 310 wide at 390x844 against the
 * contract's 370, because 40+40 of root padding had been taken out of it.
 *
 * `margin-inline: auto` RATHER THAN A LAYOUT, for the same reason: it is what centres a column
 * whose width is its own, and core's flow layout centres nothing.
 */
.wm-news-article {
	width: calc(100% - 30px);
	margin-inline: auto;
}

@media (max-width: 639.98px) {
	.wm-news-article {
		width: calc(100% - 20px);
	}
}

@media (min-width: 1200px) {
	.wm-news-article {
		max-width: 970px;
	}
}

/*
 * .wm-product-detail: THE ARTICLE'S OWN WIDTH, at every captured viewport, for the identical
 * reason `.wm-news-article` above needs its own rule rather than a block attribute.
 *
 * WHAT THE CONTRACT MEASURES (units/product-detail.yml, the root `core/group`'s own
 * `configuration.why`): "Measured root -- 1440x900: 1170x2862 at x=135; 1024x768: 994x3799;
 * 768x1024: 738x3828; 390x844: 370x6507." Read as gutters against the viewport: 1440-1170=270
 * (135 a side, i.e. the box stops growing at 1170 and centres); 1024-994=30 (15 a side);
 * 768-738=30 (15 a side); 390-370=20 (10 a side).
 *
 * WHY NOT THE `layout.constrained` / `contentSize:1170px` ALREADY ON THIS GROUP IN
 * templates/single-product.html. That attribute is real and is kept -- it is what caps and
 * centres the group's own CHILDREN (the head columns, the post-content, the related-products
 * pattern) at 1170 -- but per `.wm-news-article`'s own comment above, a constrained layout's
 * `contentSize` caps the CHILDREN of the block that declares it, never the block itself. With
 * no rule here the article's own box is simply `main`'s full width, because `main` in this
 * template is `layout:{type:"default"}` (flow, no cap of its own). MEASURED ON THIS INSTALL
 * BEFORE THIS RULE: `.wm-product-detail` rendered 1440 wide at the 1440 viewport where its own
 * contract measures 1170 -- exactly the shape this comment predicts, confirmed rather than
 * assumed.
 *
 * THE ARITHMETIC IS A FLAT 1170px CAP WITH A FLUID GUTTER BELOW IT, NOT
 * `.wm-news-article`'s TWO-GUTTER Foundation SHAPE. `calc(100% - 30px)` capped by
 * `max-width:1170px` reproduces all three of the numbers above with no second breakpoint:
 * at 1440 the fluid width (1410) exceeds the cap, so `max-width` wins and the box is 1170,
 * centred at x=135; at 1024 and 768 the fluid width (994, 738) is under the cap, so it applies
 * directly. Only the mobile gutter narrows, at this theme's own established 639.98px edge,
 * the same one `.wm-news-article` and the header/search rules above already key to.
 */
.wp-block-group.wm-product-detail {
	width: calc(100% - 30px);
	/* Corrected 2026-09-22: was a flat 1170px, which missed the same 1124-1299.98px step every
	   other Foundation-row-derived cap in this theme needed (theme.json's `styles.css` escape
	   hatch, commit 997252e). Confirmed directly against source (`#block-washington-mills-content
	   > article`, probe.js): 1070px at 1124/1200/1299, 1170px at 1300/1440. */
	max-width: calc(var(--wp--style--global--content-size) - 30px);
	margin-inline: auto;
	/* The constrained layout adds theme.json's 15px root padding (`has-global-padding`); the source's
	   article is the 1170 box itself with its children flush to both edges (measured 135..1305 at 1440). */
	padding-inline: 0;
}

/*
 * .wm-product-detail, content links and the document link -- measured on the live source
 * (/products/brown-fused-aluminum-oxide, 1440x900, 2026-09-21, probe.js + a computed-style walk):
 *   - links inside the hero blurb and the body: rgb(219, 36, 11) at rest, rgb(188, 31, 9) on hover, no
 *     underline either state (the hover value is theme.json's `elements.link` :hover, restated here
 *     because a colour set with this rule's specificity would otherwise beat that :where()-wrapped rule);
 *   - the document link (`.documents-item a`): 18px / 700, uppercase, primary red, a file glyph before and a
 *     chevron after, hover the same darker red. The glyphs are an icon font on the source and `::before`/
 *     `::after` masks here, coloured by `currentColor` so they follow the hover.
 * Focus is the browser default on both sides (outline auto 1px, offset 1px) so nothing is written for it.
 */
.wm-product-detail .wm-page-hero__content a:not(.wp-element-button),
.wm-product-detail .wp-block-post-content td a:not(.wp-element-button) {
	color: var(--wp--preset--color--primary);
}

.wm-product-detail .wm-page-hero__content a:not(.wp-element-button):hover,
.wm-product-detail .wp-block-post-content td a:not(.wp-element-button):hover {
	color: #bc1f09;
}

.wm-product-detail .wp-block-file:not(.wp-element-button) {
	margin-block: var(--wp--preset--spacing--25) 0;
	font-size: 18px;
	line-height: 27px;
}

.wm-product-detail .wp-block-file > a:not(.wp-block-file__button) {
	display: inline-block;
	position: relative;
	font-weight: 700;
	text-transform: uppercase;
	color: var(--wp--preset--color--primary);
	padding-right: var(--wp--preset--spacing--40);
	/* Lets a long, unbroken title wrap instead of overflowing its column. */
	max-width: 100%;
	overflow-wrap: break-word;
}

.wm-product-detail .wp-block-file > a:not(.wp-block-file__button):hover {
	color: #bc1f09;
}

.wm-product-detail .wp-block-file > a:not(.wp-block-file__button)::before,
.wm-product-detail .wp-block-file > a:not(.wp-block-file__button)::after {
	content: "";
	display: inline-block;
	background-color: currentColor;
	vertical-align: -3px;
	mask-repeat: no-repeat;
	mask-position: center;
	mask-size: contain;
}

.wm-product-detail .wp-block-file > a:not(.wp-block-file__button)::before {
	width: 1em;
	height: 1.15em;
	margin-right: 0.4em;
	mask-image: url( "data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 16 20'%3E%3Cpath fill='none' stroke='%23000' stroke-width='1.5' d='M1 1h9l5 5v13H1zM10 1v5h5'/%3E%3C/svg%3E" );
}

.wm-product-detail .wp-block-file > a:not(.wp-block-file__button)::after {
	width: 0.6em;
	height: 0.8em;
	margin-left: 0.5em;
	mask-image: url( "data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 10 16'%3E%3Cpath fill='none' stroke='%23000' stroke-width='2.5' d='M2 1l7 7-7 7'/%3E%3C/svg%3E" );
}

/*
 * A single product image is drawn at its own pixel size, not stretched to the column. Source (measured 2026-09-21,
 * /products/alumina-zirconia/norzon-nv-1620/ at 1440): the 470px column holds a 357x281 image at 357x281; core's
 * `.wp-block-post-featured-image img { width: 100% }` scaled it to 470x370 and pushed the body 90px down. `max-width: 100%`
 * (also core's) still keeps a larger image inside the column.
 */
.wm-product-detail__image .wp-block-post-featured-image img {
	width: auto;
}

@media (max-width: 639.98px) {
	.wp-block-group.wm-product-detail {
		width: calc(100% - 20px);
	}
}

/*
 * `.wm-products-listing__grid` NEVER RELIES ON `core/post-template`'s OWN on-demand stylesheet
 * FOR `list-style: none`, ONLY ON THIS RULE -- found 2026-09-22 while fixing the AJAX
 * filter/load-more bug below: `core/post-template`'s registered style
 * (`wp-block-post-template-inline-css`, the ONLY place core itself carries `list-style: none` for
 * this block) is missing from a page whose first paint had ZERO results (confirmed: present in a
 * normal /products/ page's <head>, absent from a filtered URL with no matches -- WordPress's
 * on-demand core-block-asset loading did not enqueue it for that request, and re-deriving exactly
 * why goes deeper into core's asset-loading internals than this fix needs). The Interactivity API
 * store (inc/products-listing.php, products-listing-view.js) BUILDS this grid client-side from
 * scratch when a later filter change DOES have results, carrying the same classes core would
 * generate -- so this one universal, always-true declaration is written directly into a
 * stylesheet this page always loads (`enqueue_block_assets`, unconditional -- see
 * functions.php's washington_mills_enqueue_styles()), rather than trusting core's own on-demand
 * loading for a grid this theme depends on either way.
 */
.wm-products-listing__grid {
	list-style: none;
}

/*
 * `products-listing` -- THE EXPOSED FILTER ABOVE THE PRODUCT GRID.
 *
 * The unit root is the `core/query` in templates/archive-product.html (`.wm-products-listing`);
 * the form it holds is a `core/group` with `tagName: form`, which is why these rules live in this
 * file. The label/select/button rules are descendants of that group (their markup is in the
 * `core/html` inside it), so core/group's presence is still the correct load condition and one
 * unit's rules stay in one place instead of being split across group.css and a new html.css for
 * three declarations.
 *
 * WHAT THE CONTRACT MEASURES, and it is the only part of this unit its capture measured at all --
 * the payload reports `{measured: 15, present: 666}`, so not one product card has a box in it:
 *
 *   form.views-exposed-form   1200x93 at x=120   (viewport 1440x900)
 *   four div.js-form-item     240x93  at x=133, 398, 663, 929
 *   label                     240x32, 18px
 *   select.form-select        240x45
 *
 * THE 25px `blockGap` WAS THE CONTRACT'S OWN ARITHMETIC, written as a block attribute, not
 * here: 398 - (133 + 240) = 25, and 663 - (398 + 240) = 25. The third gap measures 26
 * (929 - (663 + 240)) and nothing is written for the 1px -- four 240px items and three 25px gaps
 * is 1035px inside a 1170px content box, so the row does not fill its line and the residual is
 * sub-pixel rounding rather than a fourth gap value. SUPERSEDED 2026-09-25: snapped to
 * `spacing|30` (20px), spacing-standard.md §1 item 3 -- templates/archive-product.html now
 * writes that preset, not the measured 25.
 *
 * WHY THERE IS NO HORIZONTAL INSET ON THE FORM, which is the one place two of the contract's
 * numbers had to be reconciled. The `core/query` is constrained to `contentSize: 1170px`, so at
 * 1440 its content box is 1170 at x=135 -- but the contract measures the FORM at 1200 wide at
 * x=120. Those are the same box read at two edges: the source is Foundation 6, whose row gutter
 * is 0.9375rem = 15px a side, and 1200 - 30 = 1170 with 120 + 15 = 135. So the form's outer box
 * is the 1200 and its content edge is the 1170 the block attribute already asserts; adding a
 * 13px padding here to drive the items to the measured x=133 would be counting that gutter twice.
 * A 2px residual against the measured 133 is left standing rather than padded away.
 *
 * THE 16px BETWEEN LABEL AND SELECT IS DERIVED, NOT MEASURED: 93 - 32 - 45 = 16, and the
 * projection carries no y position for either child, so whether those 16 pixels sit between the
 * two or below the select cannot be read from it. Taken as a gap, which makes the item's content
 * exactly its measured 93 with no trailing margin. Named in the build hand-back.
 *
 * WHAT IS DELIBERATELY NOT WRITTEN: the select's border, fill, text colour and typography. The
 * capture recorded boxes only -- no `--props` reading reached this contract -- so any of those
 * would be invention. The selects render in the browser's own control styling until a
 * measurement says otherwise.
 *
 * THE SUBMIT'S TREATMENT IS DIFFERENT FROM THE SELECTS' AND IS NOT A MEASUREMENT EITHER --
 * it is the operator's own confirmation, recorded in `ai/units.yml`'s checkpoint for this unit:
 * "the button is not styled at all ... the build's renders as a bare default button", against
 * this theme's own established filled-CTA treatment (a "real CTA", the operator's words). See
 * the rule below the submit's own alignment rule for what was done and why.
 */
/*
 * PASS 4 (2026-09-20) -- THE FILTER'S BOX AND TYPE, RE-MEASURED ON THE LIVE SOURCE
 * (probe.js --props on https://www.washingtonmills-com.l5.wfdev.net/products, 1440x900), which
 * supersedes the passes above where they disagree. What the earlier passes could not measure and
 * this one did:
 *
 * - THE FORM IS 1200 WIDE AT x=120 WITH ITS FIRST ITEM AT x=133, not 1170 at x=135. Four 240px
 *   items, four 25px gaps and the 113px Apply are 1173px, three more than a 1170 row holds, which
 *   is what wrapped the button. The query root is already 1200 at x=120; core's constrained layout caps
 *   every child of it at the 1170 `contentSize` and centres it with `margin-inline: auto
 *   !important`, hence the !important and the two-class selector, which must beat core's
 *   `:where()` rule on specificity. The form lifts that cap back to 1200 and takes 13px of left
 *   padding, which lands the first item at 120 + 13 = 133. The 16px margin on the actions item
 *   needs the same two-class reach: `.is-layout-flex > :is(*, div){margin:0}` is (0,1,1).
 * - LABEL AND SELECT TEXT ARE 18px / 700 / UPPERCASE / black, Jura. The label is 32.4px tall
 *   (line-height 1.8) and the select sits directly under it: the item's 93px is 32.4 + 45 + a
 *   16px MARGIN BELOW THE SELECT, not a 16px gap between the two, so the item's gap is 0 and the
 *   margin is on the control.
 * - THE SELECT is white with a 1px solid rgb(172,172,172) border, padding 0 24px 0 8px,
 *   `appearance: none` and a 9x6 dark-grey triangle drawn as a background image, which is
 *   Foundation's own select rule -- reproduced with its own SVG rather than a token, since the
 *   colour lives inside the image.
 * - APPLY is 18px/700 uppercase, padding 13px 30px, 113x44, rgb(219,36,11) on white text. Its
 *   `::after` plate is computed on the source but never painted -- the source control is an
 *   `<input type=submit>`, a replaced element -- so no plate is drawn here either. It sits with
 *   its bottom margin (16) flush with the row's bottom, so `align-self: flex-end` plus that margin
 *   reproduces its measured y (316, one pixel under the select's 315).
 */
.wm-products-listing > .wm-products-listing__filter {
	width: 100%;
	max-width: 1200px;
	margin-inline: 0 !important;
	padding-left: 13px;
}

.wm-products-listing__filter-item {
	display: flex;
	flex-direction: column;
	gap: 0;
	width: 240px;
}

.wm-products-listing__filter-label {
	display: block;
	font-size: 18px;
	font-weight: 700;
	line-height: 1.8;
	text-transform: uppercase;
}

.wm-products-listing__filter-select {
	width: 100%;
	height: 45px;
	margin: 0 0 var( --wp--preset--spacing--25 );
	/* Was 0 24px 0 8px (Foundation's own select padding, reproduced): 24 snapped to
	   spacing|30 (20px), 8 to spacing|20 (10px), 2026-09-25 (spacing-standard.md §1
	   item 3, "standard wins") -- a small, visible departure from the exact Foundation
	   reproduction. */
	padding: 0 var(--wp--preset--spacing--30) 0 var(--wp--preset--spacing--20);
	border: 1px solid rgb(172, 172, 172);
	border-radius: 0;
	background-color: #fff;
	background-image: url("data:image/svg+xml;utf8,<svg xmlns='http://www.w3.org/2000/svg' version='1.1' width='32' height='24' viewBox='0 0 32 24'><polygon points='0,0 32,0 16,24' style='fill: rgb(70, 70, 70)'></polygon></svg>");
	background-origin: content-box;
	background-position: calc(100% + 16px) 50%;
	background-repeat: no-repeat;
	background-size: 9px 6px;
	color: #000;
	font-family: inherit;
	font-size: 18px;
	font-weight: 700;
	line-height: 1.5;
	text-transform: uppercase;
	-webkit-appearance: none;
	appearance: none;
}

.wm-products-listing > .wm-products-listing__filter > .wm-products-listing__filter-actions {
	align-self: flex-end;
	margin-bottom: var( --wp--preset--spacing--25 );
}

/*
 * THE SUBMIT IS THIS THEME'S OWN PRIMARY CTA TREATMENT, REACHED THROUGH THE SAME TOKENS EVERY
 * OTHER FILLED BUTTON ON THE SITE USES -- NOT A NEW STYLE.
 *
 * `wp-element-button` IS ADDED TO THE BUTTON'S CLASS LIST IN templates/archive-product.html.
 * `WP_Theme_JSON::ELEMENTS` compiles `styles.elements.button` to
 * `:where(.wp-element-button, .wp-block-button__link)` (reference/sources/block-attributes.md
 * §WP_Theme_JSON::ELEMENTS), so that one class is what makes a plain `<button>` resolve
 * theme.json's button element -- background `primary`, text `white`, weight 700, uppercase,
 * radius 0 -- with no value restated here. The mechanism already runs elsewhere on this exact
 * install: the header's search submit (`core/search`, `buttonPosition: button-outside`) renders
 * a native `<button class="wp-block-search__button wp-element-button">` and was measured on
 * this install computing the same background/color/weight/case/radius this rule relies on.
 *
 * ONE PROPERTY IS OVERRIDDEN, AND IT IS NOT A TOKEN CHANGE. theme.json's button padding is
 * `20px 0 20px 30px` -- right is deliberately 0 -- because every OTHER filled CTA this theme has
 * built pairs it with an offset fill-plate `::after` (assets/css/blocks/button.css) that paints
 * 30px of extra background to the right, so the label ends up visually inset from a painted box
 * wider than the anchor's own. That plate is a chrome row on THOSE units' own contracts, measured
 * against the source's `%cta-button` mechanism; this control has no such row, is not a marketing
 * CTA, and inventing a plate for it here would be adding chrome nothing asked for or measured.
 * Left alone, the anchor's own 0 right padding reads as text jammed against the box's right edge.
 * `padding-right` is set to the SAME preset the left side already uses --
 * `var(--wp--preset--spacing--40)`, not a new number -- so the box is symmetric with no plate.
 *
 * `cursor: pointer` IS THE ONE DECLARATION HERE THAT IS NOT FROM A TOKEN. Measured on this
 * install before this rule: an unstyled `<button>` computes `cursor: default`, and neither
 * layer of theme.json sets `cursor` on the button element. Every other clickable CTA on the site
 * is an anchor, which is a pointer by browser default with nothing written for it; a submit
 * button needs the declaration written because there is no anchor default to inherit.
 *
 * THE SELECTOR CARRIES TWO CLASSES AND NOT ONE, AND THAT IS NOT STYLE -- A ONE-CLASS RULE HERE
 * LOSES THE TIE. `global-styles-inline-css` compiles `styles.elements.button` to
 * `:root :where(.wp-element-button, .wp-block-button__link){ … padding-right: 0; … }` --
 * `:root` contributes one class-equivalent and `:where()` contributes nothing, so that rule is
 * specificity (0,1,0), and it is enqueued AFTER this file (`washington-mills-block-group-css`
 * prints before `global-styles-inline-css` in this install's own `<head>`, confirmed by reading
 * the served page). A bare `.wm-products-listing__filter-submit` is also (0,1,0): equal
 * specificity resolves by source order, and the later rule -- the global one -- would win,
 * silently reinstating `padding-right: 0`. Measured on this install before the second class was
 * added: with a one-class selector the button still computed `padding: 20px 0px 20px 30px`, the
 * override having no effect. `reference/sources/wordpress-core.md` §Global styles emit
 * `border-width: 0` for `.wp-element-button` records the same tie on a different property and
 * the same fix: a second class most specific to this instance, `.wp-element-button`, reaching
 * (0,2,0), which beats (0,1,0) regardless of source order.
 *
 * WHAT IS STILL NOT DONE. No hover state is written, because none is established for a filled
 * CTA anywhere else in this theme either (button.css's hover skew is explicitly "NOT
 * REPRODUCED" on every unit that carries it). Whether taxonomy filtering can work end to end at
 * this rung is untouched -- a separate, already-flagged finding, not this rule's question.
 *
 * PASS 3, VERTICAL PADDING: THE SAME OVERRIDE MECHANISM AS `padding-right` ABOVE, APPLIED TO
 * `padding-top`/`padding-bottom`, BECAUSE THE INHERITED BUTTON ELEMENT MADE THIS CONTROL
 * TALLER THAN THE ROW IT SITS IN. theme.json's button padding is `20px 0 20px 30px` --
 * `spacing-30` (20px) top and bottom -- and with the button's own content (this theme's
 * button typography, unchanged: 21px/700/uppercase, 31.5px line-height) that computed a
 * 71.5px box (122x72 painted) against the row's four `select.wm-products-listing__filter-select`
 * at a measured 45px. All measured on this install, at 1440x900, before this pass:
 * `.wm-products-listing__filter` (the row) is 93px tall at y=167; each label is 32px, so the
 * label/gap boundary sits at y=199 and the select's own top edge at y=215 (199 + the 16px
 * gap the block comment above derives). `.wm-products-listing__filter-actions`'s
 * `align-self: flex-end` (above, unchanged) bottom-anchors the button, so its 27px of excess
 * over the select's 45px (72-45) came entirely off the TOP: the button's top edge sat at
 * y=189 -- inside the label's own 167-199 span, not merely close to it.
 *
 * `spacing-20` (10px) REPLACES `spacing-30` ON TOP AND BOTTOM BECAUSE IT IS THE SMALLEST
 * NAMED PRESET THIS THEME DECLARES -- `settings.spacing.spacingSizes` bottoms out there --
 * so the box shrinks using a token the theme already defines rather than an invented pixel
 * value or a change to the button's type. Measured on this install after, same viewport:
 * 31.5px content plus 10 and 10 is a 51.5px box, 122x52 painted -- 6.5px of excess over the
 * select rather than 27, so the button's top edge (still bottom-anchored) moves to y=209 --
 * past the label's y=199 bottom edge and 6px short of the select's own y=215 top, i.e. inside
 * the 16px label-to-select gap rather than inside the label's own span. THE ~7px OF EXCESS IS
 * LEFT STANDING: there is no smaller named preset to close it with, and closing it exactly
 * would mean either an invented number or a font-size change this rule has no measurement to
 * justify.
 *
 * NOTHING ELSE ON THIS RULE MOVED -- `padding-right`, `cursor`, the fill, text colour,
 * weight, case and radius are pass 1's and pass 2's own declarations, and
 * `.wm-products-listing__filter-actions`'s `align-self: flex-end` and the row-fit
 * `margin-left` media rule above are both untouched.
 *
 * SUPERSEDED 2026-09-25: top/bottom padding (13px, drifted from the 10px this comment's own
 * arithmetic reasoned to) is now `spacing|25` (16px, the theme's new step), and left/right
 * (already on-scale at 30px) is now that same preset reference rather than a literal --
 * spacing-standard.md §1 item 3. The row-height arithmetic above is historical, not current.
 */
.wm-products-listing__filter-submit.wp-element-button {
	padding: var( --wp--preset--spacing--25 ) var( --wp--preset--spacing--40 );
	font-size: 18px;
	line-height: 1;
	cursor: pointer;
}

/*
 * THE LOADING INDICATOR (operator report, 2026-09-22): no visual feedback existed while a
 * filter/load-more fetch was in flight. `[hidden]` is toggled by `data-wp-bind--hidden` in
 * inc/products-listing.php against `context.isLoading`, which fetchProducts() in
 * products-listing-view.js already sets for the duration of every request -- no new state, one
 * small spinner in the theme's own primary red plus a screen-reader-text label the `aria-live`
 * wrapper announces. Centred under the grid/pager the same way the filter row above it is: no
 * new layout primitive, this unit's existing flex/centring conventions.
 */
.wm-products-listing__loading {
	display: flex;
	align-items: center;
	justify-content: center;
	gap: var(--wp--preset--spacing--20);
	width: 100%;
	padding: var(--wp--preset--spacing--30) 0;
}

.wm-products-listing__loading[hidden] {
	display: none;
}

.wm-products-listing__spinner {
	width: 24px;
	height: 24px;
	border: 3px solid var(--wp--preset--color--border-gray, #acacac);
	border-top-color: var(--wp--preset--color--primary);
	border-radius: 50%;
	animation: wm-products-listing-spin 0.8s linear infinite;
}

@media (prefers-reduced-motion: reduce) {
	.wm-products-listing__spinner {
		animation-duration: 2.4s;
	}
}

@keyframes wm-products-listing-spin {
	to {
		transform: rotate(360deg);
	}
}

/*
 * .wm-product-teaser-card: THE CARD'S HOVER DISCLOSURE, AND THE PANEL THAT CENTRES ITS CTA.
 *
 * THESE THREE RULES WERE MISFILED IN assets/css/blocks/paragraph.css AND ARE NOW WHERE THEY
 * BELONG. The file they came from said so itself and said why: "The panel collapse, the panel
 * reveal and the featured-image slide have `core/group` and `core/post-featured-image` as their
 * subjects ... They are here because this unit always places a `core/paragraph` -- the CTA -- so
 * paragraph.css is always loaded wherever the disclosure exists. MOVE THEM when group.css is
 * free; the selectors do not change."
 *
 * THE MOVE IS NO LONGER TIDINESS -- IT IS WHAT KEEPS THE DISCLOSURE ALIVE. The CTA is now a
 * `core/read-more`, so this unit places no `core/paragraph` at all and `wp_enqueue_block_style`
 * would never load paragraph.css on a page whose only paragraphs are the ones the product-data
 * block renders from PHP (that map is keyed on block type, not on emitted tags). Left where they
 * were, the panel would stop collapsing -- with no error, and looking like a layout bug.
 *
 * THE FEATURED-IMAGE RULE IS FILED BY ITS SCOPE ROOT AND NOT BY ITS TARGET, which is a knowing
 * deviation from this theme's one-file-per-block convention. Its subject is
 * `core/post-featured-image` and a post-featured-image.css does not exist; standing one up for a
 * single two-declaration rule costs a file plus a row in functions.php's map, on a file the
 * dispatch names as contended. It is scoped to `.wm-product-teaser-card`, a `core/group` hook
 * class, and the card cannot render without its root group, so this file always loads where the
 * rule is needed. Raised in the hand-back rather than left silent.
 *
 * THE DISCLOSURE IS CSS AND NOT THE INTERACTIVITY API. The contract carries no `behaviour` key at
 * all -- a discovery gap, named in the hand-back -- but its `states.hovered` entry quotes the
 * source's own rule from scss/modules/_views.scss:156-168, and its `chrome` rows quote the
 * at-rest half from :142-146 and :161-163 with a probe beside it: `.expand-content` is
 * `overflow: hidden; max-height: 0` at rest and `max-height: 400px` on `:hover`, and
 * `.top-content` goes to `margin-top: -220px; opacity: 0`. That is a `:hover` selector and a
 * `max-height`. There is no state to hold, nothing to hydrate and nothing to mirror on the
 * server, so there is no Interactivity API module here and nothing for `data-wp-*` to do.
 *
 * `text-align: center` IS THE ONE DECLARATION ADDED RATHER THAN MOVED, and it replaces a
 * mechanism the CTA used to carry itself. As a `core/paragraph` the CTA wore
 * `has-text-align-center`; `core/read-more` renders a bare `<a>` and `core/group` declares no
 * `typography.textAlign` support (block.json), so there is no attribute to write and the centring
 * has to be a rule on the panel. IT IS DERIVED FROM THE MEASUREMENT, not chosen: the contract
 * measures the anchor 178 wide at x=231 inside a 310-wide panel at x=165 -- 66px of gap on each
 * side -- and the only mechanism that centres an INLINE box is its containing block's
 * `text-align`. So the source's own `.expand-content` is centred, and the `product-data` rows the
 * panel also holds are centred in the source for the same reason.
 *
 * THAT INHERITANCE IS UNMEASURED ON BOTH SIDES AND IS FLAGGED. No artifact on either side has
 * ever rendered `product-data` inside this card -- the placement was a `core/pattern` reference
 * to a slug nothing registers, so it emitted zero bytes until this pass -- and the unit's own
 * capture is of a card whose panel is collapsed. The deduction above is the grounds; a capture of
 * the hovered card is what would confirm it, and `states.hovered` already records that no such
 * capture was taken.
 *
 * `.bottom-content`'s HALF OF THE SOURCE RULE IS NOT WRITTEN. The contract does not model that
 * element -- it "measures 310x0 at y=706 with no children and no painted style" -- so its
 * `max-height: 500px` at rest and `0` on hover have no box to act on.
 *
 * NO TRANSITION IS WRITTEN AND THE SOURCE CERTAINLY HAS ONE. The projection carries no duration,
 * easing or property list for either the panel or the image, and a plausible `0.3s ease` is an
 * invented value. Named in the hand-back as the measurement that would close it; the reveal works
 * without it.
 */
/*
 * .wm-product-teaser-card: THE CARD'S OWN ROOT IS CENTRED IN THE SOURCE, AND THIS BUILD WAS
 * MISSING IT. Added 2026-09-13, caught by `verify-unit`'s own root-level comparison rather than
 * by eye: both instance pages measured `textAlign: start` on `:root` against a reference `center`.
 *
 * CODE: `.view-mode__teaser_card.products { text-align: center; }`
 * (scss/modules/_views.scss:116-120) is the source's OWN card-root rule -- the SAME rule this
 * unit's post-title and bottom-content nodes already cite as their reason for writing an explicit
 * `textAlign: center` attribute, and the same rule `.wm-product-teaser-card__expand`'s own rule
 * below was already carrying for its own descendants. What was missing was the rule ON THE ROOT
 * ITSELF: `core/group` declares no top-level `textAlign` support (block.json), so there is no
 * attribute this card's own root could carry it as, the same reason post-title and the two
 * bottom-content nodes each carry their own explicit override instead of inheriting one.
 *
 * NOT REDUNDANT WITH THE PER-NODE OVERRIDES ALREADY ON post-title AND THE TWO bottom-content
 * NODES: those three are unaffected either way, since each already asserts its own `textAlign`.
 * This rule is what centres anything on this card that does NOT carry its own override -- the
 * product-data rows nested inside `.wm-product-teaser-card__expand`, none of which assert a
 * textAlign of their own (product-data has no ambient parent styling to inherit from on its own,
 * standalone instance page, which is a separate, already-flagged finding about that page's own
 * wrapper -- see the build hand-back -- but on THIS card, where product-data is actually and
 * normally used, this rule is what supplies the centring the reference shows).
 */
.wm-product-teaser-card {
	text-align: center;
}

/*
 * HOVER/FOCUS EXPANSION -- REWORKED TWICE ON 2026-09-21; THE SECOND IS THE RULE THAT STANDS.
 * Operator: "all these expand effects on hover need to do what is best practice and not what source had.
 * so they should probably expand but not overflow the container card". A DECIDED DEVIATION FROM THE SOURCE.
 *
 * MEASURED on the live source (Playwright, hover vs rest, 1440/1124/768/390): the card is `height: 500px;
 * overflow: hidden`, its layout box is 500px in both states and no sibling moves; `.expand-content` goes
 * `max-height 0 -> 400px` inside that box, so a card with more populated rows than fit is CLIPPED in the source.
 * The build differed: it was `min-height: 500px`, so the panel grew the article (503px, then 599px with six
 * rows) and pushed or overlaid the row below.
 *
 * THE RULE HERE: the article is a fixed 500px column and never grows. The panel is the flexible child
 * (`flex: 0 1 auto; min-height: 0`), so what it can reveal is whatever the title leaves of the 500px, and
 * anything taller is compacted (five or more populated rows drop to 16px type and a 4px gap) and, as the last
 * resort, scrolls INSIDE the panel with a thin visible scrollbar -- never clipped silently, never overflowing.
 * Empty rows are hidden (the source omits an empty field's wrapper). `:focus-within` opens the same state as
 * `:hover`: the title link and the CTA are both tab stops. Touch has no reveal, as in the source.
 *
 * GATED `@media (min-width: 640px)` 2026-09-25 (Direction A) -- this whole fixed-column mechanism is a
 * desktop/tablet concept now. The approved plan drops the fixed 500px height below 640px in favour of a
 * two-column grid with no hover state at all ("PRODUCT CARD DIRECTION A -- MOBILE", further down this file);
 * left ungated, `height: 500px` would still be a fixed height on mobile, exactly what the plan forbids.
 */
@media (min-width: 640px) {
	.wm-product-teaser-card {
		display: flex;
		flex-direction: column;
		height: 500px;
		overflow: hidden;
	}

	.wm-product-teaser-card > * {
		flex: 0 0 auto;
	}
}

/* GATED `@media (min-width: 640px)` 2026-09-25 (Direction A), same reason as the block above: below 640px the
 * panel is always open with no clipping and no reveal (its own mobile rule, further down this file). */
@media (min-width: 640px) {
	.wm-product-teaser-card__expand {
		overflow: hidden;
		max-height: 0;
		text-align: center;
	}

	.wm-product-teaser-card:hover .wm-product-teaser-card__expand,
	.wm-product-teaser-card:focus-within .wm-product-teaser-card__expand {
		flex: 0 1 auto;
		min-height: 0;
		max-height: 40rem;

		/*
		 * SCROLL ONLY ONCE SETTLED (2026-09-21, operator: a scrollbar flashed while the panel grew). During the 0.5s
		 * max-height transition the content is taller than the panel, so `overflow-y: auto` painted a classic
		 * scrollbar for ~250ms. The panel stays `hidden` while it animates; a zero-length animation delayed by the
		 * transition's 0.5s flips it to `auto`, which only draws a bar if the content still overflows the settled
		 * panel. Un-hover drops the animation, so it snaps back to `hidden` at once.
		 */
		overflow-y: hidden;
		/* Firefox: the only engine `scrollbar-width`/`scrollbar-color` reach (no `::-webkit-scrollbar`
		 * support). Firefox's own `thin` variant genuinely renders with no arrow-button decoration, so
		 * this pair is a correct, sufficient answer there on its own. */
		scrollbar-width: thin;
		scrollbar-color: var(--wp--preset--color--primary) transparent;
		animation: wm-product-teaser-card-settle 0s 0.5s forwards;
	}
}

/*
 * WEBKIT/BLINK SCROLLBAR CHROME -- ADDED 2026-09-22 (operator screenshot: "stupid scrollbars in
 * the cards on hover ... a fat/wide red bar with visible up/down triangle arrow buttons"). The
 * rule above this one is Firefox-only: `scrollbar-width`/`scrollbar-color` are not implemented by
 * Chromium or Safari at all (MDN, both properties), so on those engines the panel fell through to
 * the platform's own default scrollbar renderer -- which is exactly a classic, full-width bar with
 * arrow-button end caps on the operator's browser. Chromium/WebKit's own answer is the
 * `::-webkit-scrollbar*` pseudo-element family, which takes full control of the affordance rather
 * than tuning the native one: an element that declares `::-webkit-scrollbar` paints ONLY what
 * these rules describe -- no arrow buttons are drawn because none is one of the parts this family
 * can draw (`::-webkit-scrollbar-button` is the only part that could add one, and nothing here
 * declares it) -- so a `width` this small with a plain thumb/track is arrow-free by construction,
 * not by suppression. Unconditional on `.wm-product-teaser-card__expand` itself rather than nested
 * under `:hover`/`:focus-within`: the pseudo-elements only ever paint while the box is actually
 * scrolling (`overflow-y: auto` AND `scrollHeight > clientHeight`), which is already gated by the
 * `hidden` -> `auto` transition above, so scoping this to the hover selector too would be
 * redundant, not safer.
 */
.wm-product-teaser-card__expand::-webkit-scrollbar {
	width: 6px;
}

.wm-product-teaser-card__expand::-webkit-scrollbar-track {
	background: transparent;
}

.wm-product-teaser-card__expand::-webkit-scrollbar-thumb {
	background-color: var(--wp--preset--color--primary);
	border-radius: 3px;
}

@keyframes wm-product-teaser-card-settle {
	to {
		overflow-y: auto;
	}
}

/*
 * IMAGELESS CARD -- A DECIDED DEVIATION FROM THE SOURCE, 2026-09-21 (operator: option B). 32 products have no featured
 * image. MEASURED on the live source (1440x900, /products/alumina-zirconia/, ZF Alundum): the source keeps the image
 * slot as a ~340px blank `.top-content`, so the title sits 265px down the card at rest and jumps to 47px on hover.
 * This build's `core/post-featured-image` renders nothing without an image, so the title collapsed to 45px from the
 * top with the lower half of the 500px tile blank -- the operator called that broken. The blank hole is a source
 * defect and is NOT reproduced; instead the text block (title, classification, teaser) is centred VERTICALLY in the
 * fixed 500px tile. On hover the flexible reveal panel takes the free space, the block stays centred as it grows and
 * the title glides up to the top edge -- inside the card, no sibling moves. No placeholder photograph is invented.
 * Image cards are untouched: `:not(:has(.wp-block-post-featured-image))` never matches them.
 *
 * GATED `@media (min-width: 640px)` 2026-09-25 (Direction A) -- `justify-content` is a flex-column concept
 * (the fixed-height mechanism above, also gated) and does nothing on the mobile grid.
 */
@media (min-width: 640px) {
	.wm-product-teaser-card:not(:has(.wp-block-post-featured-image)) {
		justify-content: center;
	}
}

/*
 * The thumbnail fills the card's content width (310px at 1440), as on the source, where a 600x430 derivative is scaled to it
 * (measured 310x222 at 1440). The figure had shrunk to the imported file's own width (293, 262 ...), so cards showed
 * images of different widths. Height stays the block's 220px (source 222.2), which the hover slide below is written against.
 *
 * UNGATED, UNLIKE THE HOVER SLIDE BELOW: `width: 100%` is correct at both breakpoints -- it fills the card's
 * content column at >= 640px and the 110px thumbnail column below it (Direction A's mobile rule, further
 * down this file sets the column's own track width; this rule just fills whatever column it lands in).
 */
.wm-product-teaser-card .wp-block-post-featured-image {
	width: 100%;
}

/*
 * GATED `@media (min-width: 640px)` 2026-09-25 (Direction A) -- "hover and focus change nothing" below 640px
 * (the mobile thumbnail stays put, always visible); ungated, `:focus-within` alone (no real hover on touch)
 * would still fire this on a mobile keyboard tab into the card.
 */
@media (min-width: 640px) {
	.wm-product-teaser-card:hover .wp-block-post-featured-image,
	.wm-product-teaser-card:focus-within .wp-block-post-featured-image {
		margin-top: -220px;
		opacity: 0;
	}
}

.wm-product-data__row:has(> p:empty) {
	display: none;
}

/*
 * "FIVE OR MORE POPULATED ROWS: TIGHTER REVEAL" -- REMOVED 2026-09-25 (Direction A), SUPERSEDED
 * RATHER THAN LEFT IN PLACE. This 2026-09-21 special case shrank the rows to 1rem/1.3 line-height
 * and zeroed their gap once a fifth row was populated, so a six-row panel would still fit the
 * fixed-height card without scrolling. Direction A's compact rows ("PRODUCT CARD DIRECTION A",
 * further down this file) are UNCONDITIONALLY 16px/1.35 with a zero gap and a 1px border doing the
 * row separation instead of whitespace -- smaller and tighter than this rule ever made the old
 * 18px/20px stacked rows, at every row count, not only at five or more. Keeping both would have
 * left two selectors of comparable specificity fighting over the same properties for no visible
 * difference; the scroll fallback on `__expand` (above) is the remaining, still-real last resort
 * for a value long enough to wrap past even the compact row height.
 */

/*
 * .wm-product-teaser-card__bottom: THE TEASER COPY, THE EXACT INVERSE OF `__expand` ABOVE --
 * VISIBLE AT REST, HIDDEN ON HOVER. Added 2026-09-13 alongside the bottom-content group itself
 * (patterns/product-teaser-card.php), once content-model/product.yml's `field_teaser_subheader`/
 * `field_teaser_text` were read as this unit's own declared readers.
 *
 * CODE: scss/modules/_views.scss:147-150 sets `.bottom-content { overflow: hidden; max-height:
 * 500px; transition: all 0.3s ease-in-out; }` at rest -- 500px is far above the measured 60px, so
 * nothing is visually clipped here the way `.expand-content` clips its own contents; the rule
 * only primes the transition. On `:hover` (scss/modules/_views.scss:164-168, already cited on
 * this unit's own `states.hovered` entry), the SAME element goes to `opacity: 0; max-height: 0` --
 * so the teaser copy is exactly what a visitor sees on load and exactly what disappears when
 * `.expand-content` opens.
 *
 * "NO TRANSITION IS WRITTEN" IS STALE, LEFT FOR THE TRAIL: a `transition: all 0.3s ease-in-out`
 * for this selector was added later (CARD HOVER, further down this file), which is also the exact
 * duration Direction A now asks for -- nothing to change here.
 *
 * GATED `@media (min-width: 640px)` 2026-09-25 (Direction A) -- the mobile teaser copy is always
 * visible, in its own grid cell, with no fade (its own mobile rule, further down this file).
 */
@media (min-width: 640px) {
	.wm-product-teaser-card__bottom {
		overflow: hidden;
		max-height: 500px;
	}

	.wm-product-teaser-card:hover .wm-product-teaser-card__bottom,
	.wm-product-teaser-card:focus-within .wm-product-teaser-card__bottom {
		opacity: 0;
		max-height: 0;
	}
}

/*
 * .wm-hero-wrapper: THE ONE DECLARATION THAT MAKES THE BREADCRUMB TRAIL LAND ON THE HERO.
 *
 * THE SOURCE'S OWN REGION, measured against the source at /about-washington-mills on 2026-09-11
 * and handed to this build as a verification-side reading:
 *
 *     div.hero-wrapper
 *       |- section#block-washington-mills-breadcrumbs > nav    335x86 @120,117
 *       `- .view-mode__hero.landing_page                      1440x207 @0,117
 *
 * Both children start at the same y. The trail is `position: absolute` -- the `breadcrumbs`
 * contract records that as a measured computed value at all four viewports -- and an absolutely
 * positioned box is laid out against its nearest POSITIONED ancestor, so the wrapper is what turns
 * the trail's `top: 0` into "the top of the photograph" rather than "wherever it happened to sit".
 * Without it the trail keeps its static position and stacks above the hero, which is the defect
 * this rule closes.
 *
 * WHY THE COVER CANNOT BE THAT ANCESTOR BY ITSELF, and it looks as though it should be: core
 * declares `.wp-block-cover { position: relative }` AND `.wp-block-cover .wp-block-cover__inner-container
 * { position: relative }` (wp-includes/blocks/cover/style.css, read on this install). Every inner
 * block of a cover is serialised inside that inner container, so a trail placed there would be
 * positioned against the inner container -- which `contentPosition: bottom center` bottom-aligns
 * inside the cover's content box, 80px of padding below the cover's top edge and lower still
 * whenever the content is shorter than the box. A sibling in a wrapper is the arrangement that has
 * a stable zero, and it is also the arrangement the source uses.
 *
 * NOTHING ELSE IS DECLARED HERE. The wrapper is `layout: {type: default}` so it adds no global
 * padding and no `blockGap`, and its only in-flow child is the cover, so it takes the cover's own
 * height and paints nothing. The 1440-wide bleed is `align: full` on the block, not a rule.
 *
 * `color` IS THE SECOND DECLARATION AND IT IS THE SOURCE'S OWN, READ OUT OF THE CODE CHANNEL.
 * `page-hero`'s contract quotes the source's SCSS for this exact element -- `.hero-wrapper { color:
 * $white }` -- as the independent confirmation of its own `textColor: white` on the h1. The region
 * is where the source sets the colour, so the region is where this build sets it, and that is why
 * `breadcrumbs`' contract records a colour for the current item and NOTHING for the Home link: the
 * link never had a colour of its own on either side, it inherited one from here.
 *
 * WHAT IT CHANGES, named because it is a computed value the comparator reads: the trail's root
 * computes `color: rgb(255, 255, 255)` where it computed `rgb(0, 0, 0)` before this rule. That is
 * the other half of the defect this package was sent at -- "the built trail reads dark-on-white
 * while the source's is white-on-photograph". `.wm-page-hero` itself is unaffected: core already
 * gives a dimmed cover `color: #fff`, measured on this install before this rule.
 *
 * ITS NODE IS IN patterns/page-hero.php AND IN NEITHER CONTRACT. `page-hero`'s `structure` carries
 * no region for the trail and `breadcrumbs`' contract carries no `requires` naming `page-hero`.
 * Both are wp-spec's to record; both are named in the hand-back.
 */
.wm-hero-wrapper {
	position: relative;
	color: var(--wp--preset--color--white);
}

/*
 * `tall-image-nav-row` -- SUPERSEDED 2026-09-26, ai/card-grids-plan.md Option B. The band's
 * ground is now this block's own colour SUPPORT (`backgroundColor` / `style.color.background`,
 * any palette colour) rather than the retired `field_theme_options`, and its "Textured" choice
 * is a registered block STYLE (functions.php `washington_mills_block_styles()`) painted by the
 * identical `.wp-block-group.is-style-textured` rule `core/group`'s own registered "Textured"
 * style already carries above -- this unit's rendered root IS a `core/group`
 * (blocks/tall-image-nav-row/render.php), so no `.wm-tall-image-nav-row`-scoped rule is needed
 * here any more. `gray` and `textured` still have no measurement anywhere in this unit's
 * projection; named in the hand-back, same finding as before this rewrite.
 */

/* -------------------------------------------------------------------------- header
 *
 * Unit: header (parts/header.html). Hook class .wm-header on the part's root group.
 */

/*
 * THE HEADER IS A POSITIONED, RAISED BOX -- ADDED 2026-09-13, AND IT IS THE SOURCE'S OWN RULE
 * RATHER THAN A MECHANISM INVENTED FOR THE SEARCH PANEL. CODE, quoted in this unit's contract
 * on the root node: `scss/layout/_header.scss:3-9` -- `header.page-header { ... box-shadow: 0 0
 * 8px rgba($black, 0.2); position: relative; z-index: 5 }`. The box-shadow half of that rule is
 * already carried as the root group's `style.shadow` attribute; these two declarations had
 * nowhere to go, because core/group's own `position` support is gated to the theme opting into
 * it and neither value has an ordinary attribute at all. The contract records them as a
 * "checked-and-inert reading" for exactly that reason.
 *
 * THEY STOPPED BEING INERT THE MOMENT THE SEARCH PANEL WAS BUILT. `states.search-open` measures
 * the panel as a full-width band BELOW the header, so the only thing it can be anchored to is
 * the header's own box -- which needs `position: relative` to be a containing block -- and it
 * has to paint OVER the page content that follows it, which is what `z-index: 5` buys on a
 * header that comes first in document order. Both values are the source's, not fitted here.
 */
.wm-header {
	position: relative;
	z-index: 5;
}

/*
 * BELOW THE SOURCE'S `large` EDGE THE HEADER IS 80px TALL, NOT 106 -- ADDED 2026-09-20. MEASURED
 * (Playwright, computed, source vs build, 1024 / 768 / 390): source header 80 / 80 / 67 with
 * root padding 15px 0 and a logo 321x50 at x=15 (240x37 at 390, scaling with its column, which is
 * `small-8` = 66.67% there); the build's was 106 / 106 / 94 -- 30px block padding at every width,
 * 15px of extra root side padding (the theme's own `has-global-padding`, which the source header
 * does not have) and a logo pinned at 294 by its block attribute. The edge is 1123.98px, the
 * same `large` boundary the collapse in navigation.css is keyed to. `!important` on the two block
 * paddings is because they are the block's own inline style (the 30px is the contract's
 * `spacing|40` at 1440, which is correct there); an inline value is beaten by nothing else.
 */
@media (max-width: 1123.98px) {
	.wp-block-group.wm-header {
		/* Was 15px top/bottom (matching the source's own root padding 15px 0, a vertical value,
		   not the shared left/right token): snapped to spacing|25 (16px), 2026-09-25
		   (spacing-standard.md §1 item 3). */
		padding-top: var(--wp--preset--spacing--25) !important;
		padding-bottom: var(--wp--preset--spacing--25) !important;
		padding-left: 0;
		padding-right: 0;
	}

	.wm-header .wp-block-column:first-child {
		flex-basis: 66.67% !important;
	}

	.wm-header .wp-block-site-logo img {
		width: 321px;
		max-width: 100%;
		height: auto;
	}
}

/* At the source's small step the column gutter is 10px rather than 15 (measured: logo x=10,
   240 wide in a 260px column at 390; x=15 from 640 up). The value is the block's own inline
   padding, hence !important. */
/*
 * THE HEADER ROW IS 1100px WIDE BELOW THE SOURCE'S `xlarge` STEP AND 1200px FROM IT -- ADDED
 * 2026-09-21, PROMOTED TO A SITE-WIDE TOKEN THE SAME DAY.
 *
 * MEASURED on the live source (Playwright, computed, resized in place): `header .row` is 1100px
 * wide and centred at every width from 1124 to 1299 (x=12 at 1124, 40 at 1180, 50 at 1200, 75 at
 * 1250, 90 at 1280, 99.5 at 1299) and 1200px wide from 1300 (x=50 at 1300, 120 at 1440).
 *
 * THIS IS NOT THE HEADER'S OWN NUMBER. `.row{max-width:75rem}` (1200px) with
 * `@media(max-width:81.24875em){.row{max-width:68.75rem}}` (1100px, edge 1299.98px) is served
 * from `/themes/washington_mills/css/washington_mills.css` and applies to every `.row`-classed
 * container on the source -- header, hero, breadcrumbs, content sections, carousels alike, all one
 * Foundation rule. Confirmed directly: `probe.js` against the same live source measured
 * `#block-washington-mills-breadcrumbs` (also a bare `.row`, no header markup at all) at the same
 * 1100/1124-1299 and 1200/>=1300 split, independent of anything header-specific. So this was a
 * shared token wearing a per-unit patch, not a header fact -- CLAUDE.md "A defect present in more
 * than one unit is fixed once, at its root." theme.json's `styles.css` now carries
 * `--wp--style--global--content-size`/`--wp--style--global--wide-size: 1100px` inside its existing
 * `@media (width <= 1299.98px)` block (the same edge already used there for `intro-text`), which
 * `core/group`'s own `.is-layout-constrained > :where(…)` rule already reads as `max-width` --
 * confirmed against this install's own `global-styles-inline-css` output, not assumed from memory.
 * That is the general fix now; every constrained container narrows to 1100px in this band with no
 * per-unit CSS.
 *
 * REPLACES THE EARLIER breadcrumbs.css NOTE, WRITTEN THE SAME DAY, WHICH SAID THIS BAND WAS
 * "1070px... between 1124 and 1319px." That number and that upper edge were never measured against
 * the source; this comment's own figures (1100px, edge 1300px, all individually measured on the
 * live source, corroborated by two independent instruments -- the served stylesheet's own compiled
 * media query and a live probe of a second, unrelated `.row`) supersede it.
 *
 * WHAT STAYS HEADER-ONLY, AND WHY. `--wp--style--global--content-size` resolves through
 * `max-width`, which never forces a box wider than its available space -- unlike the literal
 * `width: 1100px` this rule used to also write, which is what made the padding-zero below
 * necessary in the first place. Measured (probe.js, before/after, `.wm-header > .wp-block-columns`,
 * every 1px-relevant width 1024-1440): with the root fix ALONE and this rule fully removed, the
 * header reproduces the source's own cited x-values EXACTLY from 1130px up (1100/1200 widths,
 * 15/25/40/50/75/90/100/50/120 as the source measures) -- except 1124-1129px, a 6px-wide sliver,
 * where `--wp--style--root--padding-left/right` (15px each, `useRootPaddingAwareAlignments`) eats
 * into the 1100px cap and the header measures 1094-1099px instead: 1124 - 30 = 1094 < 1100. The
 * source's own `.row` carries no side padding of its own at all (established already for
 * breadcrumbs: "no inner padding"), so this collision is WordPress's root-padding-aware-alignments
 * feature meeting a source that has none -- not a mismeasurement, and not something the generic
 * content-size token can express (it does not touch a full-width block's own padding). None of
 * this toolkit's four standard captured viewports (1440/1024/768/390) fall inside 1124-1129, so
 * the un-padded version is invisible to every existing comparison; zeroed, it is exact everywhere
 * measured. Kept narrow and named, not generalised: zeroing every full-width block's root padding
 * site-wide in this band was not checked against any OTHER unit and is not asserted here.
 */
@media (min-width: 1124px) and (max-width: 1129.98px) {
	.wp-block-group.wm-header {
		padding-left: 0;
		padding-right: 0;
	}
}

@media (max-width: 639.98px) {
	.wm-header .wp-block-column {
		padding-left: var(--wp--style--root--padding-left, 10px) !important;
		padding-right: var(--wp--style--root--padding-right, 10px) !important;
	}
}

/*
 * LINK-DESCRIPTION CARD BACKGROUND (2026-09-22) -- .wm-link-description-cards__card. MEASURED on
 * the live source (Playwright, computed-style walk of `.paragraph--type--link-description-card`,
 * the individual card element -- NOT `.link-card-wrapper`/`.link-card`, the row/column wrapper,
 * which really is transparent and is a different node): `background-color: rgb(255, 255, 255)`,
 * `box-shadow: rgba(0, 0, 0, 0.2) 0px 0px 7px 0px`. This card previously carried neither, which read
 * as text floating on the band with no boundary; `ai/units/link-description-cards.yml` carried the
 * same error against the wrong node, corrected the same day -- see that file's own core/column
 * `why`. The shadow value is the same as `.wm-news-teaser-card` / `.wm-product-teaser-card`'s own
 * rest-state shadow (WHOLE-CARD LINK, below), but those two carry it as an inline `style.shadow`
 * block attribute already, which is what forces `!important` on THEIR hover rule; this card has no
 * such inline attribute, so a plain rule is enough here and needs no `!important`. It does not fight
 * the hover rule directly below (CARD HOVER), which only ever changes `transform` on this selector,
 * nor the stretched-link overlay (WHOLE-CARD LINK, below), whose `::after` paints no background of
 * its own.
 */
.wm-link-description-cards__card {
	background-color: rgb(255, 255, 255);
	box-shadow: rgba(0, 0, 0, 0.2) 0px 0px 7px 0px;
}

/*
 * LINK-DESCRIPTION CARD EQUAL HEIGHT (2026-09-22) -- .wm-link-description-cards__row >
 * .wp-block-column. MEASURED on the live source (Playwright, getBoundingClientRect() on
 * `.paragraph--type--link-description-card`, 1440x900): every row's two cards share EXACTLY the
 * same height -- 206.8px for four of the five rows, 233.8px for the "Case Studies"/"FAQ's" row,
 * 260.8px for the lone ninth card -- never a per-card natural height. The source's own mechanism
 * (`.link-card.columns`, the column-equivalent) is `display: flex` around its single child, so
 * `align-items: normal` (== `stretch`) fills that child to the column's own height, and the COLUMN
 * itself is stretched in turn by `.link-card-wrapper.row`'s own flex-wrap line. Measured build
 * BEFORE this rule: `.wp-block-column` (core's own CSS -- wp-includes/blocks/columns/style.css)
 * IS a stretched flex item of `.wp-block-columns` already, but is not itself a flex/grid container
 * for ITS OWN child, so `.wm-link-description-cards__card` -- a nested `core/group` -- only ever
 * took its own natural content height and the column's stretch went nowhere visible: "Case Studies"
 * 270.4px next to "FAQ's" 238.9px in the same row, a 31.5px gap, same pattern on
 * "Literature"/"ISO Certifications". `display: grid` on the column is enough on its own -- CSS
 * Grid's initial `align-items`/`justify-items` are both `stretch`, so the single grid item (the
 * card) fills the column's full height AND width with no extra rule needed, and does not disturb
 * the column's own flex-item sizing within `.wp-block-columns` (that is controlled by flex-grow /
 * flex-basis, orthogonal to how the column lays out its own children). CONFIRMED AT 390px: the
 * pattern's `core/columns` stacks single-column below 782px (`isStackedOnMobile: true`), so each
 * row holds one card with no sibling to stretch against -- measured byte-identical before and
 * after this rule at 390x844, so it is not scoped to a media query.
 */
.wm-link-description-cards__row > .wp-block-column {
	display: grid;
}

/*
 * CARD HOVER -- link-description, news-teaser and product-teaser. MEASURED on the live source (Playwright,
 * 1440x900, hover vs rest, 2026-09-21); CODE for each rule is named beside it.
 *
 * .wm-news-teaser-card and .wm-product-teaser-card: `.card-view .views-row .view-mode__teaser_card:hover`
 * is `background-color: #e1e2e2; box-shadow: 0 0 1px rgba(0,0,0,.2); transform: scale(1.05)`, measured as
 * rgb(225, 226, 226), the 7px shadow collapsing to 1px, and matrix(1.05, ...) at both /news and /products,
 * under `transition: background-color 0.5s ease-in-out, transform 0.5s ease-in-out`. The product card's
 * own expand/collapse (above) is the same hover; the source times it `all 0.5s ease-in-out` on
 * `.top-content` and `.expand-content` and `all 0.3s ease-in-out` on `.bottom-content`, measured as those
 * computed transitions.
 *
 * `transform: scale(1.05)` IS NO LONGER WRITTEN AT ALL, ON EITHER CARD -- A KNOWING DEVIATION, NOT A GATE
 * (changed 2026-09-24, superseding the 2026-09-23 `@media (min-width: 1124px)` gate below, which itself
 * superseded the ungated original). Operator, relayed by the main session, ai/inventory.yml
 * selection.site_wide: "can we also fix these stupid expanding cards so they don't grow, so the inside
 * expands and then crops at the card borders and not making the card grow stupidly? this is one of the
 * bs design features of the source site that is terrible and not best practice." Read as: no scale on
 * the card, so its size and position never change on hover, on any viewport; whatever expands inside
 * stays within the card's border and is cropped by it; the reveal behaviour (background-color,
 * box-shadow, and the product card's own `__expand` panel above) stays exactly as measured. The
 * 2026-09-23 gate is superseded rather than deleted-and-forgotten: it fixed a real full-bleed overflow at
 * 390px (a card at rest `x:10, width:370` became `x:0.75, width:388.5` scaled), but this instruction
 * removes the cause outright rather than hiding it below 1124px, so the gate's own reasoning is now moot
 * on both cards -- there is nothing left to overflow. `background-color` is unaffected and stays
 * unconditional; `box-shadow` is NOT -- see the dated paragraph directly below this one, which
 * supersedes the "7px shadow collapsing to 1px" reading above for the hover rule itself (the reading is
 * kept because it is still what the live source does; the build now deliberately does not). `.wm-link-
 * description-cards__card:hover` (its own `scale(1.05)`, gated at the theme's `large`/70.25em edge,
 * immediately below) and `.wm-navigation-card:hover` (`scale(1.1)` in post-template.css) are NOT part of
 * this instruction and are untouched -- see the CARD HOVER TRACKING note below this rule.
 *
 * `box-shadow: 0 0 1px rgba(0,0,0,.2)` ON HOVER IS NO LONGER WRITTEN EITHER, ON EITHER CARD -- A SECOND
 * KNOWING DEVIATION FROM THE SAME SOURCE RULE, ADDED 2026-09-24. Operator, relayed by the main session,
 * ai/inventory.yml selection.site_wide, 2026-09-24: "the card keeps its rest shadow (`0 0 7px
 * rgba(0,0,0,.2)`) in every state, so it never looks bigger" -- direction A, approved against
 * ai/tmp/design-audit/mock-card-a-grey-1440-hover.png, where the hovered card's edge is the same as its
 * neighbours' at rest. The source's own hover rule REPLACES the 7px rest shadow with a thinner 1px one,
 * which reads as the card looking smaller/flatter on hover rather than more raised; deleting the
 * `box-shadow` declaration from this rule (previously `0 0 1px rgba(0,0,0,.2) !important`) leaves the
 * card's own inline `style.shadow` attribute (`0px 0px 7px 0px rgba(0,0,0,.2)`, both patterns' root
 * group) as the only source of the property, so it now reads identically at rest and on hover/focus-
 * within on both cards. The `!important` the old rule needed to outrank that inline value is gone with
 * it -- nothing here needs to beat the inline style any more, it simply leaves it alone.
 */
.wm-news-teaser-card,
.wm-product-teaser-card {
	transition: background-color 0.5s ease-in-out;
}

.wm-news-teaser-card:hover,
.wm-news-teaser-card:focus-within,
.wm-product-teaser-card:hover,
.wm-product-teaser-card:focus-within {
	background-color: #e1e2e2;
}

/*
 * CARD HOVER TRACKING NOTE (2026-09-24) -- two OTHER cards still scale on hover and were reported, not
 * fixed, because the operator's instruction above named only the product and news teaser cards:
 * `.wm-link-description-cards__card:hover` (`scale(1.05)`, below) and `.wm-navigation-card:hover`
 * (`scale(1.1)`, post-template.css). If a future instruction extends the "no growing cards" decision to
 * either of them, this is where the same fix goes.
 */

/*
 * 0.5s -> 0.3s, 2026-09-25 (Direction A): "transitions are 0.3s" (ai/status.md 2026-09-24) for the image
 * slide and the panel reveal -- the teaser copy (`__bottom`, below) was already 0.3s and needed no change.
 * `.wm-news-teaser-card` shares none of these three selectors, so its own reveal stays 0.5s; Direction A's
 * "News card: the shadow fix only" names no duration change there, and none is made here.
 */
.wm-product-teaser-card .wp-block-post-featured-image,
.wm-product-teaser-card__expand {
	transition: all 0.3s ease-in-out;
}

.wm-product-teaser-card__bottom {
	transition: all 0.3s ease-in-out;
}

/*
 * BACKGROUND-COLOR TO 0.3s, PRODUCT CARD ONLY, 2026-09-25 (Direction A). The shared `background-color 0.5s`
 * rule above (CARD HOVER) is `.wm-news-teaser-card, .wm-product-teaser-card` together, and Direction A's
 * 0.3s instruction is this card's alone -- rewriting the shared selector would also retime the news card's
 * fill, which nothing here was asked to touch (CLAUDE.md: a single-unit difference stays that unit's fix
 * until a second unit adopts it). Same specificity as the shared rule (one class), later in the cascade, so
 * it wins for `.wm-product-teaser-card` without needing `!important` -- `background-color` is not an inline
 * style on this element, only `box-shadow` and the four paddings are.
 */
.wm-product-teaser-card {
	transition: background-color 0.3s ease-in-out;
}

@media (min-width: 1124px) {
	.wm-link-description-cards__card {
		transition: transform 0.3s ease-in-out;
	}

	.wm-link-description-cards__card:hover {
		transform: scale(1.05);
	}
}

/*
 * PRODUCT CARD DIRECTION A (2026-09-25) -- .wm-product-teaser-card, the rest of the operator's approved
 * plan not already covered by the gates and the transition/background-color rules above. Full plan text:
 * ai/status.md 2026-09-24 "Operator decisions ... product cards direction A"; carried into the contract as
 * ai/units/product-teaser-card.yml's root `configuration.why`, "THE APPROVED PLAN, 2026-09-24 (DIRECTION
 * A)". Mockups: ai/tmp/design-audit/mock-card-a-grey-1440-hover.png, mock-card-mobile-390-rest.png -- read
 * for the design, not build input; every value they were drawn with is written out below and in the
 * contract, per that file's own header.
 */

/*
 * TITLE MARGIN WHILE THE PANEL IS OPEN -- item 2, "the title has 20px below it (var:preset|spacing|30)".
 * `!important` because post-title's own `style.spacing.margin.bottom: 9px` (patterns/product-teaser-
 * card.php) is an inline style; nothing else beats it. Gated to >= 640px: below it there is no hover state
 * at all (item 5).
 */
@media (min-width: 640px) {
	.wm-product-teaser-card:hover .wp-block-post-title,
	.wm-product-teaser-card:focus-within .wp-block-post-title {
		margin-bottom: var(--wp--preset--spacing--30) !important;
		transition: margin-bottom 0.3s ease-in-out;
	}
}

/*
 * SPACE UNDER THE TABLE, BEFORE THE CTA -- item 4, "20px under the table, then the CTA". Scoped to a
 * direct child of THIS card's own expand panel, so a standalone `product-data` instance (product-detail,
 * related-products) is untouched -- its own contract is unchanged, per the `core/pattern` node's own `why`
 * in the contract.
 *
 * UNGATED 2026-09-25 (ONE FIX PASS, item 5) -- was `@media (min-width: 640px)` only, because the mobile
 * rule had no CTA and the approved plan's own item 5 said "no 20px under the table" there. The operator's
 * checkpoint superseded that sentence: below 640px "Learn More" is now shown, in the desktop reveal's own
 * style -- which includes this same 20px of space above it (read-more.css, mobile section). Ungating this
 * rule rather than duplicating it at both widths.
 */
.wm-product-teaser-card__expand > .wm-product-data {
	margin-bottom: var(--wp--preset--spacing--30);
}

/*
 * COMPACT SPEC ROWS, THIS CARD ONLY -- item 4. `product-data`'s own contract and its own
 * `patterns/product-data.php` are UNCHANGED; this is the "rule scoped to this card" the contract's own
 * `core/pattern` node already names as the placement's compromise. Values are the approved plan's, not a
 * measurement: label left in a 47% column with 12px after it, value in the rest; both 16px/1.35, body
 * font; label `dark-gray` (#464646) weight 400, value `black` weight 700; each row 6px padding above/below
 * its text with a `border-gray` (#acacac) line above it, the last row also below; no gap between rows;
 * left-aligned (the rest of the card stays centred -- `text-align: left` is written here, not inherited,
 * because `.wm-product-teaser-card { text-align: center }`, above, would otherwise cascade onto these rows
 * the way it does onto product-data's other placements).
 *
 * SPECIFICITY, NOT `!important`: `.wm-product-teaser-card .wm-product-data__row p:first-child/last-child`
 * (0,3,1) already outranks `.wm-product-data__row p:first-child/last-child` (paragraph.css, 0,2,1 -- the
 * stacked 18px/20px default every OTHER placement of this pattern still uses); nothing here fights an
 * inline style.
 *
 * UNGATED: compact rows are the design at every width this card is placed at -- only the reveal mechanism
 * around them (clipped and hover-revealed above 640px, always open below it) differs by breakpoint, and
 * that gating is on `__expand` itself (above, and in the mobile block below), not on the rows.
 *
 * FIXED 2026-09-25 -- `gap: 0` BELOW NEVER REACHED THE ROWS' OWN SPACING, WHICH IS WHY THIS CARD
 * MEASURED 44.6px BETWEEN ROWS AGAINST THE APPROVED 34px. `wm-product-data`'s `layout.type` is
 * "default" (flow), and a flow layout spaces its children with `margin-block-start` on each row
 * after the first (from the pattern's own `blockGap`, 10px) -- CSS `gap` has no effect on a flow
 * layout at all, so this rule's `gap: 0` was a silent no-op the whole time. The real mechanism is
 * overridden directly below. Rows also snap from 6px to the theme's 10px step
 * (spacing-standard.md §1 item 3), the approved plan's number, not a re-measurement.
 */
.wm-product-teaser-card .wm-product-data {
	gap: 0;
	text-align: left;
}

.wm-product-teaser-card .wm-product-data__row + .wm-product-data__row {
	margin-block-start: 0;
}

.wm-product-teaser-card .wm-product-data__row {
	display: flex;
	align-items: baseline;
	padding-block: var(--wp--preset--spacing--20);
	border-top: 1px solid var(--wp--preset--color--border-gray);
}

.wm-product-teaser-card .wm-product-data__row:last-child {
	border-bottom: 1px solid var(--wp--preset--color--border-gray);
}

.wm-product-teaser-card .wm-product-data__row p:first-child {
	flex: 0 0 47%;
	/* Was 12px, snapped to spacing|20 (10px), 2026-09-25 (spacing-standard.md §1 item 3). */
	margin-right: var(--wp--preset--spacing--20);
	font-size: 16px;
	font-weight: 400;
	line-height: 1.35;
	color: var(--wp--preset--color--dark-gray);
}

.wm-product-teaser-card .wm-product-data__row p:last-child {
	flex: 1 1 auto;
	font-size: 16px;
	font-weight: 700;
	line-height: 1.35;
	color: var(--wp--preset--color--black);
}

/*
 * REDUCED MOTION -- item 3, "off under reduced motion". Scoped to this card's own selectors only, the same
 * set the 0.3s durations above are written on. The news card's own hover transitions are UNCHANGED and
 * still run under reduced motion: Direction A's reduced-motion instruction is this card's, named in the
 * same breath as its own 0.3s durations, and the news card was named separately in the plan ("the shadow
 * fix only"). Flagged as a related gap in the hand-back, not fixed here.
 */
@media (prefers-reduced-motion: reduce) {
	.wm-product-teaser-card,
	.wm-product-teaser-card .wp-block-post-featured-image,
	.wm-product-teaser-card__expand,
	.wm-product-teaser-card__bottom {
		transition: none;
	}

	.wm-product-teaser-card .wp-block-post-title {
		transition: none !important;
	}
}

/*
 * PRODUCT CARD DIRECTION A -- MOBILE (up to 639.98px, item 5). No fixed height, no hover state, a
 * two-column grid: a 110px image column and the rest, 20px between them. The image is top-aligned and
 * always visible, spanning both content rows; the title is first in the second column, 20px/1.25 with 4px
 * below it; the teaser copy sits under the title, always visible, both lines 16px, no margins; the compact
 * spec rows (above) run across both columns under all of that, 20px below it; there is no CTA -- the title
 * is the card's link, as it already is at every width (WHOLE-CARD LINK, below). Padding drops to 20px on
 * all sides (`var:preset|spacing|30`) from the root's own inline 30px (`var:preset|spacing|40`) --
 * `!important` because that padding is an inline style (patterns/product-teaser-card.php).
 *
 * VISUAL ORDER, NOT DOCUMENT ORDER: the markup order is image, title, expand panel, teaser-copy -- same at
 * every width -- but the approved mobile layout wants the panel AFTER the teaser copy. CSS Grid places by
 * `grid-row`/`grid-column`, which does not require DOM order to match, so the panel is given the last row
 * explicitly rather than moved in the pattern file. Flagged: the CTA is hidden here (read-more.css), but
 * the spec rows are not, so a screen reader still encounters them before the teaser copy in source order --
 * a real, if minor, reading-order question this rule does not resolve.
 *
 * `.is-layout-constrained > *` NEEDS NEUTRALISING AT EVERY WIDTH, WITH `!important` -- THE ROOT CAUSE OF
 * TWO OF THE OPERATOR'S FIVE CHECKPOINT ITEMS (ONE FIX PASS, 2026-09-25), NOT JUST A MOBILE-GRID CONCERN.
 *
 * MEASURED (Playwright, computed styles, before this fix). At 1440, card 0 hovered:
 * `.wm-product-teaser-card__expand` computed `margin-left: 66px; margin-right: 66px; width: 178px` against
 * a 310px panel -- shrink-to-fit and centred, not stretched to the panel's real width. At 390 on /products,
 * the same element's width tracked its OWN CONTENT rather than the grid track: BLASTITE's row measured
 * 230px inset ~50px each side inside a 330px content box; other cards measured 245-330px depending on their
 * data's own text length.
 *
 * THE CAUSE: `global-styles-inline-css` compiles `.is-layout-constrained` support to
 * `.is-layout-constrained > :where(:not(.alignleft):not(.alignright):not(.alignfull)){max-width:
 * var(--wp--style--global--content-size);margin-left: auto !important;margin-right: auto !important;}` --
 * `:where()` carries zero specificity, so the selector's own specificity is just one class, but the
 * `!important` on the margin wins regardless of specificity or source order. The neutraliser that existed
 * here before this pass (`margin-inline: 0`, no `!important`) never beat it, at ANY width -- it was mobile
 * -only AND a no-op even there, which is why BLASTITE's inset (item 3, checkpoint) and the wrapped spec
 * rows (item 1) were both really the same bug. `margin: auto` on a box with `width: auto` inside a flex or
 * grid container does not merely centre it -- it makes the box SHRINK-TO-FIT (the auto margins absorb the
 * free space that `stretch` would otherwise have filled), which is what narrowed the panel and wrapped its
 * text.
 *
 * THE FIX: this rule now matches `!important` for `!important` and applies UNGATED, because the card is a
 * flex column above 640px (COLUMN flex's cross axis is horizontal, same auto-margin mechanism) and a grid
 * below it -- both need it, not only the grid. Specificity is `.wm-product-teaser-card.is-layout-constrained
 * > *` = two classes (0,2,0), which already beats the `:where()`-zeroed core rule (0,1,0) on its own; the
 * `!important` is what actually matters. `margin-left`/`margin-right` (not the `margin-inline` logical
 * shorthand) are named explicitly to match the exact longhands core writes, rather than relying on the
 * browser's own logical/physical equivalence mapping.
 *
 * VERIFIED AFTER: card 0 hovered at 1440, `.wm-product-teaser-card__expand` computes `margin-left: 0px;
 * margin-right: 0px; width: 310px` (full panel width, no shrink). At 390 on /products, every card's
 * `.wm-product-data__row` (and BLASTITE's) now measures the full 330px content width with x=30, no inset.
 */
.wm-product-teaser-card.is-layout-constrained > * {
	max-width: none;
	margin-left: 0 !important;
	margin-right: 0 !important;
}

/*
 * IMAGELESS CARD, NO EMPTY 110px COLUMN -- item 4. OPERATOR DECISION AT THIS CHECKPOINT, SUPERSEDING THE
 * APPROVED PLAN'S OWN ITEM 5 SENTENCE ("A CARD WITH NO IMAGE KEEPS THE 110px COLUMN EMPTY"): with no image
 * the text takes the full width, no empty column and no gap reserved for it. Cards with an image are
 * unchanged -- this selector only ever matches the imageless case.
 *
 * `core/post-featured-image` RENDERS NOTHING AT ALL WHEN THE PRODUCT HAS NO FEATURED IMAGE -- not an empty
 * wrapper, no element in the DOM (measured: the instance page's `.wp-block-post-featured-image` query
 * returns null for Alumina-Zirconia). So `grid-template-columns: 110px 1fr` reserved a 110px track with
 * nothing placed in it on 3 of 24 /products cards (Alumina-Zirconia, and two more counted in the contract's
 * own root `why`). `:has()` is the same mechanism the desktop imageless rule (below) already uses for the
 * identical "no image" test, so both breakpoints answer it the same way.
 */
@media (max-width: 639.98px) {
	.wm-product-teaser-card:not(:has(.wp-block-post-featured-image)) {
		grid-template-columns: 1fr;
	}

	.wm-product-teaser-card:not(:has(.wp-block-post-featured-image)) .wp-block-post-title,
	.wm-product-teaser-card:not(:has(.wp-block-post-featured-image)) .wm-product-teaser-card__bottom {
		grid-column: 1;
	}
}

@media (max-width: 639.98px) {
	.wm-product-teaser-card {
		display: grid;
		grid-template-columns: 110px 1fr;
		column-gap: var(--wp--preset--spacing--30);
		row-gap: 0;
		align-items: start;
		text-align: left;
		padding: var(--wp--preset--spacing--30) !important;
	}

	.wm-product-teaser-card .wp-block-post-featured-image {
		grid-column: 1;
		grid-row: 1 / span 2;
	}

	.wm-product-teaser-card .wp-block-post-featured-image img {
		height: 79px !important;
	}

	.wm-product-teaser-card .wp-block-post-title {
		grid-column: 2;
		grid-row: 1;
		text-align: left !important;
		font-size: 20px !important;
		line-height: 1.25 !important;
		/* Was 4px, snapped to spacing|20 (10px, the scale's floor), 2026-09-25
		   (spacing-standard.md §1 item 3, "standard wins") -- a visible change to this
		   mobile grid card's title-to-content gap. */
		margin: 0 0 var(--wp--preset--spacing--20) !important;
	}

	.wm-product-teaser-card__bottom {
		grid-column: 2;
		grid-row: 2;
	}

	.wm-product-teaser-card__expand {
		grid-column: 1 / -1;
		grid-row: 3;
		margin-top: var(--wp--preset--spacing--30);
		max-height: none;
		overflow: visible;
		text-align: left;
	}

	/* "Hover and focus change nothing at this width" -- item 5: neutralises the shared CARD HOVER fill
	   (above) for this card only, on mobile only. .wm-news-teaser-card is untouched. */
	.wm-product-teaser-card:hover,
	.wm-product-teaser-card:focus-within {
		background-color: transparent;
	}
}

/*
 * PAGE HERO VARIANTS (2026-09-21) -- rules for patterns/page-hero-{photo-left,microsite-band,band-plain}.php.
 * Every value below was read off the live source with a computed-style walk of `.hero-wrapper` at 1440x900
 * and 390x844 on 2026-09-21 (contact-us, documents, fused-minerals/fused-mineral-industries); the
 * measurements are in each pattern's header. Nothing here repaints a group that is not inside a
 * `.wm-page-hero` or `.wm-hero-wrapper`.
 *
 * `--light`: the wrapper's `color: white` above is right for the two bands that sit over a dark ground
 * (image, microsite) and wrong where the ground is white -- the current breadcrumb item computes black on
 * contact-us and white on about-washington-mills, and it inherits the wrapper's colour.
 */
.wm-hero-wrapper--light,
.wm-page-hero--photo-left {
	color: var(--wp--preset--color--black);
}

/*
 * photo-left geometry, both steps computed on the source's `.inner-hero-wrapper` (contact-us): 1440 -> row padding
 * 50px 0 60px, columns 0 15px, text column 10px 15px 0 30px, blurb padding-left 40px; 390 -> row 30px 0, columns
 * 0 10px, text column 30px 10px 0, blurb 0. The step is placed at the theme's mobile edge; the source's is
 * Foundation's 640px `medium`, the same number within a pixel. WordPress's own columns stack at 781px, so
 * 640..781 renders stacked where the source is still two-up -- UNMEASURED, and named in the hand-back.
 */
.wm-page-hero--photo-left .wp-block-columns {
	padding-top: var(--wp--preset--spacing--40);
	padding-bottom: var(--wp--preset--spacing--40);
}

.wm-page-hero--photo-left .wp-block-column {
	padding-right: var(--wp--style--root--padding-right, 10px);
	padding-left: var(--wp--style--root--padding-left, 10px);
}

.wm-page-hero--photo-left .wm-page-hero__text {
	padding-top: var(--wp--preset--spacing--40);
}

@media (width > 639.98px) {
	.wm-page-hero--photo-left .wp-block-columns {
		padding-top: var(--wp--preset--spacing--60);
		padding-bottom: var(--wp--preset--spacing--70);
	}

	.wm-page-hero--photo-left .wp-block-column {
		padding-right: var(--wp--style--root--padding-right, 15px);
		padding-left: var(--wp--style--root--padding-left, 15px);
	}

	.wm-page-hero--photo-left .wm-page-hero__text {
		padding-top: var(--wp--preset--spacing--20);
		padding-left: var(--wp--preset--spacing--40);
	}

	.wm-page-hero--photo-left .wm-page-hero__content {
		padding-left: var(--wp--preset--spacing--50);
	}
}

/* band-plain: the source's article is black text on the texture, while its breadcrumb inherits the wrapper's white. */
.wm-page-hero--plain {
	color: var(--wp--preset--color--black);
	background-position: 50% 50%;
}

/*
 * microsite-band. The photograph is the page's featured image, drawn as an absolute panel at the left
 * (600px wide with a 15px highlight-3 edge, z-index 3) over a dark band whose column is 700px and
 * right-aligned -- above the theme's tablet edge. At or below it (390 is the measured point) the photograph
 * fills the article behind a 70% dark band. The 1024..1123 range is UNMEASURED: the source changes at
 * Foundation's 1024 `large` and this theme's edge is 1123.98.
 */
.wm-page-hero--microsite {
	position: relative;
}

.wm-page-hero--microsite .wm-page-hero__photo {
	position: absolute;
	inset: 0;
	margin: 0;
	z-index: 0;
}

.wm-page-hero--microsite .wm-page-hero__photo img {
	width: 100%;
	height: 100%;
	object-fit: cover;
}

.wm-page-hero--microsite .wm-page-hero__band {
	position: relative;
	z-index: 1;
	background-color: rgb(from var(--wp--preset--color--dark-gray) r g b / 0.7);
}

/*
 * THE BAND'S OWN BOX IS THE FULL HEIGHT, NOT A PAINTED STRIP (2026-09-27, second operator-approved
 * pass -- supersedes the 2026-09-27 `::after` patch above, which the operator read as a visible seam:
 * "yeah but image now has gap"). `.wm-page-hero--microsite` itself carries
 * `padding-bottom:var(--wp--preset--spacing--50)` (40px) as a block attribute -- the source's own
 * `article.view-mode__hero.microsite_page { padding-bottom: 40px }` -- and that 40px was dead space:
 * `.wm-page-hero__photo` (`position:absolute; inset:0`) already spans the root's full padded height at
 * every viewport (measured on /fused-minerals/ and /silicon-carbide/, 1440x900 and 390x844: root and
 * photo both report the same box), but `.wm-page-hero__band` is in-flow and stopped 40px short of it,
 * so the photo's own last 40px showed through with no tint -- a bright strip.
 *
 * The previous fix closed that strip with a same-coloured `::after` rectangle painted on top of the
 * photo and the band. It worked as a colour match but not as an image: it replaced the photograph's own
 * pixels for that 40px with a flat fill, which reads as a seam between real photo-plus-tint above and a
 * solid block below -- reported back as "image now has gap".
 *
 * The fix instead: the root's own 40px is folded into the band's own bottom padding (both here and in
 * the `>1123.98px` block below), and the root's block attribute is set to zero
 * (patterns/page-hero-microsite-band.php). The band's real box now reaches the root's true bottom edge
 * on its own -- no second element, no repeated colour, no seam -- and the root's total height is
 * unchanged: band content + band's own padding (now including the folded 40px) equals exactly what band
 * content + band's old padding + the root's old padding-bottom summed to before. The photo's own
 * pixels, dimmed by the band's real overlay, now run the full height with nothing painted over them.
 */
.wm-page-hero--microsite .wm-page-hero__band {
	padding: var(--wp--preset--spacing--40) 0
		calc(var(--wp--preset--spacing--40) + var(--wp--preset--spacing--50));
}

.wm-page-hero--microsite .wm-page-hero__col {
	color: var(--wp--preset--color--white);
	padding: var(--wp--preset--spacing--40) var(--wp--preset--spacing--20) 0;
}

.wm-page-hero--microsite .wm-page-hero__col h1 {
	font-family: var(--wp--preset--font-family--jura);
	text-transform: uppercase;
	letter-spacing: -2px;
	line-height: 1;
	margin: 0;
}

.wm-page-hero--microsite .wm-page-hero__content {
	padding-top: var(--wp--preset--spacing--20);
}

.wm-page-hero--microsite .wm-page-hero__content p {
	margin: 0 0 var(--wp--preset--spacing--40);
	font-size: 18px;
	line-height: 1.4;
}

@media (width > 1123.98px) {
	.wm-page-hero--microsite .wm-page-hero__photo {
		right: auto;
		width: 600px;
		z-index: 3;
		border-right: 15px solid var(--wp--preset--color--highlight-3);
		box-sizing: border-box;
	}

	.wm-page-hero--microsite .wm-page-hero__band {
		background-color: var(--wp--preset--color--dark-gray);
	}

	.wm-page-hero--microsite .wm-page-hero__band {
		padding: var(--wp--preset--spacing--60) 0
			calc(var(--wp--preset--spacing--70) + var(--wp--preset--spacing--50));
	}

	.wm-page-hero--microsite .wm-page-hero__col {
		width: 700px;
		margin-left: calc(50% - 100px);
		padding: var(--wp--preset--spacing--20) var(--wp--preset--spacing--25) 0 var(--wp--preset--spacing--40);
	}

	.wm-page-hero--microsite .wm-page-hero__content {
		padding-left: var(--wp--preset--spacing--50);
	}

	.wm-page-hero--microsite .wm-page-hero__col h1 {
		font-size: 68px;
	}

	.wm-page-hero--microsite .wm-page-hero__content p {
		font-size: 26px;
		line-height: 1.4;
	}
}

/*
 * The hero blurb's own rhythm, read on the source's `.field--name-field-hero-content` (contact-us,
 * 1440x900): first paragraph flush to the h1's 8px margin, p margin 0 0 16px, h2 44px on a 1.4 line box
 * with margin 0 0 8px, links rgb(219, 36, 11). Core's paragraph and heading margins and the theme's 1.1
 * heading leading are what the block would otherwise carry.
 */
.wm-page-hero__text > .wm-page-hero__content {
	margin-block-start: var(--wp--preset--spacing--20);
}

/* The h1-to-content gap on the product-detail hero, which carries no `.wm-page-hero__text`
   wrapper of its own -- spacing-standard.md §2 "Proximity": a heading's gap to its own content,
   20px (spacing|30), added 2026-09-25. */
.wm-product-detail__text > h1 + * {
	margin-block-start: var(--wp--preset--spacing--30);
}

.wm-page-hero__content p {
	margin: 0 0 var(--wp--preset--spacing--25);
}

.wm-page-hero__content h2 {
	margin: 0 0 var(--wp--preset--spacing--20);
	line-height: 1.4;
}

/*
 * h3 GETS THE SAME TREATMENT AS h2, NOT A SECOND MEASUREMENT (2026-09-27). The field this content
 * comes from is real ACF WYSIWYG (render.php's own docblock: "paragraphs, a sub-heading, a
 * call-to-action button"), and /fused-minerals/'s own sub-heading is authored as an `<h3>`
 * ("Expertly manufactured fused mineral grains and powders", `wp post meta get 759 hero_content`)
 * where the rule above was written against a page whose field used `<h2>`. No capture measures an
 * h3 in this field specifically -- reusing the h2 rule is the same "sub-heading in the hero blurb"
 * treatment applied to the tag this page actually uses, not an invented number; named in the
 * hand-back as unverified for h3 on its own.
 */
.wm-page-hero__content h3 {
	margin: 0 0 var(--wp--preset--spacing--20);
	line-height: 1.4;
}

.wm-page-hero--photo-left .wm-page-hero__content a:not(.wp-element-button) {
	color: var(--wp--preset--color--primary);
}

/*
 * The hero's call-to-action button. On the source it is an `a.cta-button` written INSIDE an `<h2>` in the
 * hero-content field, so it inherits that h2's Roboto Slab and 1.4 line-height (computed on contact-us:
 * 21px/700 Roboto Slab, line-height 29.4px, box 284x69 at y=675 in an h2 that starts at y=670). The build's
 * button is a real core/button, so the same two declarations are written here, scoped to the hero. The source
 * wrapping a link in a heading is flagged in the hand-back (a link announced as a heading); it is reproduced
 * here for fidelity only until the operator decides.
 */
.wm-page-hero__content .wp-block-buttons {
	margin: 0 0 var(--wp--preset--spacing--20);
}

@media (width > 639.98px) {
	.wm-page-hero__content .wp-block-buttons {
		padding-top: var(--wp--preset--spacing--20);
	}
}

.wm-page-hero__content .wp-block-button__link {
	font-family: var(--wp--preset--font-family--roboto-slab);
	line-height: 1.4;
}

/* Added 2026-09-25: the product-detail hero's CTA compresses theme.json's inherited button
   padding (spacing|30, 20px top/bottom) to the theme's 16px step, spacing-standard.md §1 item 3. */
.wm-product-detail .wp-block-button__link {
	padding-top: var(--wp--preset--spacing--25);
	padding-bottom: var(--wp--preset--spacing--25);
}

/*
 * The hero's text column gutter: Foundation's `columns` padding, 10px at 390 and 15px from the theme's mobile
 * edge up (computed on the source's `.hero-wrapper .column`: 0 10px at 390x844, 0 15px at 1440x900). It is a
 * rule, not a block attribute, because a value written as an attribute is inline and no media query beats it.
 * At 390 this is what lets "About Washington Mills" (about 372px of text) sit on one line in a 370px column.
 * The plain band's h1 carries the source's 30px margin-bottom (measured on /documents at both viewports).
 */
.wm-page-hero__col {
	padding-right: var(--wp--style--root--padding-right, 10px);
	padding-left: var(--wp--style--root--padding-left, 10px);
}

@media (width > 639.98px) {
	.wm-page-hero__col {
		padding-right: var(--wp--style--root--padding-right, 15px);
		padding-left: var(--wp--style--root--padding-left, 15px);
	}
}

.wm-page-hero--plain .wm-page-hero__col h1 {
	margin-bottom: var(--wp--preset--spacing--40);
}

/*
 * THE FIVE HERO VARIANTS ADDED 2026-09-21 -- home-band, block-photo-left, block-image-band, photo-left-plain, search-band. Each
 * value was read off the live source with a computed-style walk at 1440x900 and 390x844 (and a width sweep for the steps); the
 * patterns' headers carry the readings. The `section-heading` title spec and its marker are the heading block's, in heading.css.
 */

/* home-band: the blurb's own type and the button's margin (30px above it in the source, inside an h2 so it does not collapse with the paragraph's 16px: 46px here). The source steps at this theme's edge (1123.98px). */
.wm-page-hero--home .wm-page-hero__content p {
	font-size: 20px;
	line-height: 1.6;
}

.wm-page-hero--home .wm-page-hero__content .wp-block-buttons {
	/* Was 46px 0 8px (30px source margin + the paragraph's 16px, non-collapsing): 46 snapped to
	   spacing|60 (50px), 8 to spacing|20 (10px), 2026-09-25 (spacing-standard.md §1 item 3). */
	margin: var(--wp--preset--spacing--60) 0 var(--wp--preset--spacing--20);
	padding-top: 0;
}

.wm-page-hero--home .wm-page-hero__content .wp-block-button__link {
	font-size: 18px;
}

@media (width > 1123.98px) {
	.wm-page-hero--home .wm-page-hero__content p {
		font-size: 26px;
		line-height: 1.2;
	}

	.wm-page-hero--home .wm-page-hero__content .wp-block-button__link {
		font-size: 22px;
	}
}

/* block-image-band: the title sits in the source's `medium-11` column (11/12 of the row), left aligned. Padding, not max-width,
   because core's constrained layout centres a narrowed child with an !important auto margin. */
@media (width > 639.98px) {
	.wm-page-hero--block-image .wm-page-hero__col {
		padding-right: calc(8.3333% + var(--wp--preset--spacing--25));
	}
}

/*
 * block-photo-left (`.wm-page-hero--block`): the blurb's list is the source's `ul.two-column` and carries the paragraph's own type (a bare `ul`
 * gets none of the `core/list` typography, which theme.json keys to `.wp-block-list`); the CTAs are `cta-link-button`s in a `div`, so they
 * are Jura at 1.5 where the photo-left band's `contact-us` button, written inside an `h2`, is Roboto Slab at 1.4 (source computed 2026-09-21:
 * 167x72 at 1440); the source item margin is 30px above and 50px between them (plus a 4px inline space).
 */
.wm-page-hero--block .wm-page-hero__content ul {
	margin: 0 0 var(--wp--preset--spacing--25) var(--wp--preset--spacing--30);
	padding: 0;
	font-size: var(--wp--preset--font-size--body-text);
	font-weight: 600;
	line-height: 1.6;
}

.wm-page-hero--block .wm-page-hero__content li {
	margin: 0;
}

.wm-page-hero--block .wm-page-hero__content .wp-block-buttons {
	margin: var(--wp--preset--spacing--40) 0 var(--wp--preset--spacing--20);
	padding-top: 0;
	column-gap: var(--wp--preset--spacing--60);
}

.wm-page-hero--block .wm-page-hero__content .wp-block-button__link {
	font-family: var(--wp--preset--font-family--jura);
	line-height: 1.5;
}

@media (width > 639.98px) {
	.wm-page-hero--block .wm-page-hero__content ul {
		column-count: 2;
	}
}

/*
 * photo-left-plain: a centred 8-of-12 column in a light-gray band, breadcrumb over it. Foundation's `medium` is 640px.
 * The band's own background is the `light-gray` preset on the block; only the geometry is here.
 */
.wm-page-hero--plain-gray {
	text-align: center;
}

.wm-page-hero--plain-gray .wm-page-hero__row {
	padding: var(--wp--preset--spacing--40) 0;
}

.wm-page-hero--plain-gray .wm-page-hero__col {
	padding: var(--wp--preset--spacing--40) var(--wp--preset--spacing--20) 0;
}

@media (width > 639.98px) {
	.wm-page-hero--plain-gray .wm-page-hero__row {
		padding: var(--wp--preset--spacing--60) 0 var(--wp--preset--spacing--70);
	}

	.wm-page-hero--plain-gray .wm-page-hero__col {
		max-width: 66.6667%;
		margin-inline: auto;
		padding: var(--wp--preset--spacing--20) var(--wp--preset--spacing--25) 0 var(--wp--preset--spacing--40);
	}

	.wm-page-hero--plain-gray .wm-page-hero__content {
		padding-left: var(--wp--preset--spacing--50);
	}
}

.wm-page-hero--plain-gray .wm-page-hero__col h1 {
	margin-bottom: var(--wp--preset--spacing--20);
}

/* The blurb field on /404 is raw h1 / hr / h3 markup; the source gives them margin 0 0 8, an hr of margin 20px 0 with a 1px `medium-gray` bottom border and no height, and leaves the browser's own margins nowhere. */
.wm-page-hero--plain-gray .wm-page-hero__content h1,
.wm-page-hero--plain-gray .wm-page-hero__content h3 {
	margin: 0 0 var(--wp--preset--spacing--20);
}

.wm-page-hero--plain-gray .wm-page-hero__content hr {
	height: 0;
	margin: var(--wp--preset--spacing--30) 0;
	border: 0;
	border-bottom: 1px solid var(--wp--preset--color--medium-gray);
}

/* search-band: the title is `h1.page-title` (margin 0 0 8) over the form; the 10-of-12 column starts at the source's `large` edge (1124px). */
.wm-page-hero--search .wm-page-hero__col h1 {
	margin: 0 0 var(--wp--preset--spacing--20);
}

@media (width > 1123.98px) {
	.wm-page-hero--search .wm-page-hero__col {
		max-width: 83.3333%;
		margin-inline: auto;
	}
}

/*
 * .wm-centered-text: THE BAND'S VERTICAL PADDING MOVED FROM AN ATTRIBUTE TO A RULE (2026-09-21), and the fourth
 * variant that made it necessary. patterns/centered-text.php used to write `padding-top/bottom: spacing|40`
 * (30px) as block attributes, which serialise inline, and the photo variant's padding is 154px / 160px at 1440 --
 * no class can beat an inline declaration, so the 30px had to stop being one. The gray, white and textured
 * values are unchanged: 30px, measured on about-washington-mills at both viewports.
 */
.wm-centered-text {
	padding-top: var(--wp--preset--spacing--40);
	padding-bottom: var(--wp--preset--spacing--40);
}

/*
 * THE DARK-PHOTO VARIANT. MEASURED 2026-09-21 on the live source /contact-us with a computed-style walk of
 * `.paragraph--type--centered-text.bg-image` (1440x900 and 390x844): band `position: relative`, text
 * rgb(255, 255, 255), padding 154px 0 160px at 1440 and 30px 0 at 390; the overlay is the band's `::before`,
 * `position: absolute; inset: 0; z-index: -1; background: rgba(70, 70, 70, 0.7)` -- dark-gray at 70%, so it is
 * written from the token; h2 44px/700 (28px at 390) and the intro 27px/600 (22px) are the same as the gray band's
 * and need nothing here; the unfilled "Browse Products" link is white on this band and red on the gray one.
 * `isolation: isolate` makes the group a stacking context so the z-index -1 overlay paints ABOVE the group's own
 * background photograph instead of behind it. The 154/160 step sits at the theme's tablet edge; only 1440 and 390
 * were measured, so 1024..1440 is an assumption.
 */
.wm-centered-text.is-style-dark-photo {
	position: relative;
	isolation: isolate;
	color: var(--wp--preset--color--white);
	background-size: cover;
	background-position: 50% 50%;
}

.wm-centered-text.is-style-dark-photo::before {
	content: "";
	position: absolute;
	inset: 0;
	z-index: -1;
	background-color: rgb(from var(--wp--preset--color--dark-gray) r g b / 0.7);
}

.wm-centered-text.is-style-dark-photo .wp-block-button.is-style-outline > .wp-block-button__link {
	color: var(--wp--preset--color--white);
}

/* The chevron follows the link's text colour on the source (its ::after paints no background of its own; white here, red on the gray band). */
.wm-centered-text.is-style-dark-photo .wp-block-button.is-style-outline > .wp-block-button__link::after {
	background-color: currentColor;
}

@media (width > 1123.98px) {
	.wm-centered-text.is-style-dark-photo {
		/* Was 154px/160px, clamped to the scale's top step, spacing|80 (80px), 2026-09-25
		   (spacing-standard.md §1 item 3, "standard wins") -- a large, visible reduction. */
		padding-top: var(--wp--preset--spacing--80);
		padding-bottom: var(--wp--preset--spacing--80);
	}
}

/*
 * PROXIMITY FALLBACK, 2026-09-25 (site-wide vertical-spacing correction pass,
 * ai/inventory.yml selection.site_wide 2026-09-24 entry, approved 2026-09-25).
 *
 * These two rules are NOT the primary mechanism for their gap -- the pattern files
 * (patterns/centered-text.php, patterns/inner-section-content.php) already write the
 * correct inline `margin-top` on these nodes for every FRESH instance. They exist because
 * `wp-migrate-content-import`'s stored-content rewrite (site commit 2f12cee) desynced some
 * already-imported pages' block markup from its own JSON attributes: the attribute still
 * reads `"margin":{"top":"var:preset|spacing|30"}}` but the serialised HTML lost the
 * matching `style="margin-top:..."`, so the browser renders zero. A class-based rule with
 * no `!important` only ever WINS where the inline style is missing -- it can't out-rank a
 * present inline style, so a correct instance is untouched and only a desynced one is
 * fixed. Named in the build hand-back as the stored-content shape that still needs its own
 * fix (wp-migrate-content-import owns rewriting the DB rows; this rule is the front-end
 * safety net until that runs, and stays afterwards as a defence against the same import
 * bug recurring).
 */
.wm-centered-text > .wp-block-group {
	margin-top: var(--wp--preset--spacing--30);
}

/*
 * The list-based instances of inner-section-content already carry their own inline
 * `margin-top` on `wp:list` (patterns/inner-section-content.php). Some already-imported
 * instances (/, "Control from Start to Finish" and "Products") use a plain paragraph here
 * instead of a list, with no margin of its own -- the heading and its content touch. Scoped
 * to the adjacent-sibling shape so it cannot reach any other paragraph in the theme.
 */
.wm-inner-section-content__heading + p {
	margin-top: var(--wp--preset--spacing--30);
}

/*
 * .wm-basic-content: PROXIMITY FOR A MID-CONTENT HEADING, 2026-09-25 (same pass). This band's
 * root blockGap is the theme's global default (spacing|25, 16px, patterns/basic-content.php) --
 * right for paragraph-to-paragraph flow, wrong for a heading marking a new subsection:
 * spacing-standard.md §2 "Proximity" wants 40px above a heading and 20px below it, not 16px
 * on both sides. Measured wrong on /careers/ ("Potential Careers": 16px above instead of 40;
 * "All the benefits...": 16px below instead of 20). `:not(:first-child)` leaves a heading that
 * opens the band alone -- it already gets margin-block-start:0 from core's own `:first-child`
 * layout rule, and "the gap above" there is the band's own top padding, not this rule's job.
 * Specificity (0,2,1) beats the container's own generated `.wp-container-core-group-is-layout-*
 * > * + *` rule (0,1,0) regardless of source order, so no `!important` is needed; nothing here
 * out-ranks a heading's own inline margin if a future instance ever sets one.
 */
.wm-basic-content > :is(h2, h3, h4, h5, h6):not(:first-child) {
	margin-top: var(--wp--preset--spacing--50);
}

.wm-basic-content > :is(h2, h3, h4, h5, h6) + * {
	margin-top: var(--wp--preset--spacing--30);
}

/*
 * WHOLE-CARD LINK (2026-09-21) -- .wm-news-teaser-card, .wm-product-teaser-card and
 * .wm-link-description-cards__card. Operator decision: a card that looks clickable is one link end to end.
 * MEASURED on the live source (Playwright, 1440x900): the news and product teasers are whole-card through the
 * source's own `.block-link` script (a click on the padding navigates; `elementFromPoint` there is the
 * `article`/`div`, not an `a`), the link-description card is NOT (cursor: pointer, a click on it goes nowhere --
 * a source defect, not copied). This is the stretched-link technique, so it is one anchor, one tab stop and no
 * script: the card is the containing block and the title link's own `::after` covers it. Nothing is nested --
 * the anchor is still the heading's child, as before, and the card's copy is not made interactive.
 * The keyboard ring follows the card (`outline` on the pseudo-element, in the link's own colour), not the words.
 * The hover states above and below are on the card, so they keep working: the overlay is inside it.
 */
.wm-news-teaser-card,
.wm-product-teaser-card,
.wm-link-description-cards__card {
	position: relative;
}

.wm-news-teaser-card .wp-block-post-title a::after,
.wm-product-teaser-card .wp-block-post-title a::after,
.wm-link-description-cards__card h3 a::after {
	content: "";
	inset: 0;
	position: absolute;
}

/* The stretched link's overlay makes the whole card its `:hover`, so the theme's link-hover underline painted under the
   title whenever the card was hovered; the source's own card title (`a.ref-link`) is not underlined on hover. */
.wm-news-teaser-card .wp-block-post-title a:hover,
.wm-product-teaser-card .wp-block-post-title a:hover {
	text-decoration: none;
}

.wm-news-teaser-card .wp-block-post-title a:focus-visible,
.wm-product-teaser-card .wp-block-post-title a:focus-visible,
.wm-link-description-cards__card h3 a:focus-visible {
	outline: none;
}

.wm-news-teaser-card .wp-block-post-title a:focus-visible::after,
.wm-product-teaser-card .wp-block-post-title a:focus-visible::after,
.wm-link-description-cards__card h3 a:focus-visible::after {
	outline: 2px solid currentcolor;
	outline-offset: -2px;
}

/*
 * .wm-applications-hero: THE APPLICATIONS HERO'S OWN MARGIN-BOTTOM, SCOPED SEPARATELY FROM
 * `.wm-hero-wrapper`'s BARE RULE (2026-09-24, units/applications-listing.yml,
 * patterns/applications-hero.php).
 *
 * CORRECTED 2026-09-24 (checkpoint 2, item 1) -- FLAT 50px AT EVERY WIDTH, NOT STEPPED. This
 * used to step at Foundation's `large` breakpoint (1124px) on the strength of
 * scss/layout/_hero.scss:55-59, `&.landing_page { @include breakpoint(large) { margin-bottom:
 * 50px } }` -- 0 below 1124, 50 above. That citation is real but is the WRONG hero: it is
 * `.landing_page`'s own modifier, and `.landing_page` is a page-type class the actual
 * Applications landing page carries (this theme's own page-hero-block-image-band.php, a
 * DIFFERENT pattern, was "built and measured against a flat 50px with no step" per the
 * paragraph below -- the same flat 50 this rule now uses). A term archive is not that page
 * type, so `block.block.applicationshero.yml`'s own hero-wrapper most likely never carries
 * `.landing_page` and instead renders `.hero-wrapper`'s un-modified base margin, which this
 * theme's own sibling pattern already shows is a flat 50px.
 *
 * MEASURED AT 390x844 on Vibratory Finishing (checkpoint 2, item 1): the nav
 * (`.wm-applications-menu-region`, accordion.css's own `margin-top: -75px`, shared with
 * templates/page-applications.html and NOT changed here) started at y=273 against the source's
 * y=323 -- 50px short, overlapping the hero title's last line (both end at y=288). With this
 * hero's own margin-bottom measured at 0px and the hero's box bottom at y=349, the collapsed
 * gap into the nav's -75px was -75, landing the nav 76px above the hero's own bottom edge. A
 * `margin-bottom: 50px` on the hero collapses against that same -75 to -25, landing the nav at
 * y=324 -- within a rounding pixel of the source's measured 323, and consistent with
 * page-applications.html's OWN hero (page-hero-block-image-band.php's flat 50, not the
 * `.landing_page` step) landing its own nav at y=429 against a computed hero margin-bottom of
 * ~49px there (worked out the same way: nav-top 429, hero-bottom 455, nav's own -75 margin,
 * collapsed gap -26 implies hero margin-bottom ~49). Two different hero patterns, both
 * measured to carry the same flat ~50, is why this is a flat value and not a second step.
 *
 * NOT RE-VERIFIED AT 1024, 768 OR 1440 against the source for THIS template (only 390 was
 * measured this checkpoint) -- named in the hand-back. 1440 is unaffected in outcome (50px
 * either way, stepped or flat, since 1440 is above the old rule's own 1123.98px cutoff).
 *
 * SCOPED TO ITS OWN CLASS, NOT TO `.wm-hero-wrapper` ITSELF: patterns/page-hero-block-image-band.php
 * shares `.wm-hero-wrapper` and `.wm-page-hero--block-image` for the same source hero on a
 * different page (vibratory-finishing's stray page 855) and was built and measured against a flat
 * 50px with no step -- this rule now matches it rather than diverging from it.
 *
 * TRIED STEPPED AGAIN 2026-09-25, AND REVERTED IN THE SAME PASS -- named so the next reader does
 * not retry it on the strength of the same citation. The current contract
 * (ai/units/applications-listing.yml, hero group node) carries a DIRECT measurement of the
 * source's own `section#block-applicationshero` reading margin-bottom 0 at 1024/768/390 and 50
 * only at 1440, and argues the checkpoint-2 flat value was inferred backwards from the nav's
 * landed position rather than read off the hero box itself. Rewriting this rule to step at
 * 1123.98px on that strength was measured here, on this build, at 390x844 on
 * /applications/industries/vibratory-finishing/: the hero's last title line ("Applications") sat
 * under the nav band, reproducing exactly the defect checkpoint 2 closed (nav overlapping the
 * hero title). So on THIS build, with `.wm-applications-menu-region`'s unconditional
 * `margin-top: -75px` (accordion.css) unchanged, the flat value is what avoids the overlap; a
 * source element's own isolated margin-bottom does not settle what this build's composed layout
 * needs when a sibling rule elsewhere was tuned against the old value. Left flat and reported as
 * a conflict in the build hand-back rather than decided silently either way: whether the fix
 * belongs here (this rule) or in the nav's own margin-top (a different, already-built unit,
 * applications-menu-block) is not this package's call.
 */
.wm-applications-hero {
	margin-bottom: var(--wp--preset--spacing--60);
}
