diff --git a/booth/static/embed.js b/booth/static/embed.js
index 15d2f36..2d24f96 100644
--- a/booth/static/embed.js
+++ b/booth/static/embed.js
@@ -100,6 +100,10 @@
     ".bk-ask-was{margin:.15rem 0 .55rem;font-size:.86rem;opacity:.8}",
     ".bk-ask-was b{opacity:1}",
     ".bk-ask-err{color:var(--bk-err);font-size:.86rem}",
+    /* [hidden] restated at class specificity: a host rule as plain as
+       `p{display:block}` outranks the UA's own [hidden] and would show it. */
+    ".bk-ask-status{margin:.6rem 0 0;font-size:.86rem;color:var(--bk-err)}",
+    ".bk-ask-status[hidden]{display:none}",
     "@media print{.bk-ask{break-inside:avoid}}"
   ].join("");
 
@@ -368,6 +372,133 @@
     if (e.key === "booth.theme" || e.key === null) bkTheme();
   });
 
+  /* ONE SUBMIT SAVES EVERY ASK ON THE PAGE (2026-09-27; U3 contract,
+     "Submitting several asks at once"). One pick is one <form> is one POST,
+     and that POST's 303 reload wiped every pick made in the others: the
+     operator answered top to bottom, pressed the last button, and lost the
+     rest (`auk-audition`, 15:02:23 in the access log).
+
+     A submit on one of OUR forms, while ANOTHER of ours holds input the
+     operator changed, sends every changed ("dirty") form of ours: one POST
+     each, to its own action, with `Accept: application/json` for the route's
+     204 (r2 C3), one after another in DOCUMENT ORDER of the forms. A form
+     nobody touched is not sent - re-sending re-dates an answer nobody gave, and
+     a blank one is refused - and that includes the pressed one. A refused form
+     stops nothing. Each form the server took becomes its own new baseline, so
+     it is never sent again unless it changes again.
+
+     Then the page reloads - so what shows is the server's record - ONLY if
+     nothing is left unsaved: no refusal, and nothing of ours changed while
+     the batch was in flight. Otherwise NO reload: the pressed form's status line says
+     what did not save, and everything he entered stays on the page. The page
+     stays live during the flight, which is why that second condition exists
+     (heid bug-hunt, 4 of 4 arms). A press during the flight is ignored, never
+     handed to the browser; nothing is re-sent without a fresh press.
+
+     With no batch in flight and no OTHER dirty form, this does nothing and
+     the browser submits: the plain POST and 303 it always was. */
+  var sending = false;
+
+  function serial(form) {
+    return new URLSearchParams(new FormData(form)).toString();
+  }
+
+  function dirty(form) {
+    /* Against what the server last TOOK from this page, once it has taken
+       something (`__bkSaved`, set per form on its 204). Before that, against
+       the server-rendered default: `form.elements` includes every control
+       bound by `form=`, wherever it sits in the document. */
+    if (form.__bkSaved !== undefined) return serial(form) !== form.__bkSaved;
+    var els = form.elements;
+    for (var i = 0; i < els.length; i++) {
+      var el = els[i];
+      if (el.type === "radio" || el.type === "checkbox") {
+        if (el.checked !== el.defaultChecked) return true;
+      } else if (el.tagName === "TEXTAREA" || el.type === "text") {
+        if (el.value !== el.defaultValue) return true;
+      }
+    }
+    return false;
+  }
+
+  function ourForms() {
+    /* Document order (querySelectorAll), and only forms inside something this
+       script mounted: an author's own form is theirs, never ours to send. */
+    var all = document.querySelectorAll('form[id^="bk-ask-form-"]'), out = [];
+    for (var i = 0; i < all.length; i++) {
+      for (var j = 0; j < ours.length; j++) {
+        if (ours[j].contains(all[i])) { out.push(all[i]); break; }
+      }
+    }
+    return out;
+  }
+
+  function askOf(form) {
+    var els = form.elements;
+    for (var i = 0; i < els.length; i++) {
+      if (els[i].name === "ask") return els[i].value;
+    }
+    return "?";
+  }
+
+  function send(form, body) {
+    /* Resolves to null when saved, or to why it was not. NEVER rejects, so one
+       refusal cannot stop the chain behind it. */
+    return fetch(form.action, {
+      method: "POST", body: body, credentials: "same-origin",
+      headers: { "Accept": "application/json" }
+    }).then(function (r) {
+      if (r.status === 204) return null;
+      return r.json().then(function (j) { return (j && j.detail) || "status " + r.status; },
+                           function () { return "status " + r.status; });
+    }, function () { return "the Booth did not answer"; });
+  }
+
+  document.addEventListener("submit", function (ev) {
+    var form = ev.target;
+    if (ev.defaultPrevented || !window.fetch || !window.URLSearchParams || !window.FormData) return;
+    var forms = ourForms();
+    if (forms.indexOf(form) < 0) return;
+    /* In flight FIRST: a press on a form with no other dirty form beside it
+       would otherwise fall through to the browser, a native POST racing the
+       batch (heid bug-hunt, hulda). */
+    if (sending) { ev.preventDefault(); return; }
+    var others = forms.some(function (f) { return f !== form && dirty(f); });
+    if (!others) return;                        // the browser's own POST and 303
+    ev.preventDefault();
+    sending = true;
+    var batch = forms.filter(dirty);
+    // every form's fields read NOW, at the press
+    var bodies = batch.map(function (f) { return new URLSearchParams(new FormData(f)); });
+    var failed = [], chain = Promise.resolve();
+    batch.forEach(function (f, n) {
+      chain = chain.then(function () {
+        return send(f, bodies[n]).then(function (why) {
+          if (why === null) f.__bkSaved = bodies[n].toString();
+          else failed.push(askOf(f) + " (" + why + ")");
+        });
+      });
+    });
+    chain.then(function () {
+      sending = false;                          // before anything here can throw
+      /* A refusal blocks the reload ON ITS OWN: a refused form the operator
+         then set back to its first value reads clean, and "nothing dirty"
+         alone reloaded over the failure without a word (heid bug-hunt,
+         hulda). Otherwise anything dirty was touched mid-flight. */
+      var changed = forms.some(dirty);
+      if (!failed.length && !changed) { location.reload(); return; }
+      var saved = batch.length - failed.length;
+      var st = form.parentNode && form.parentNode.querySelector(".bk-ask-status");
+      if (!st) return;
+      st.textContent = failed.length
+        ? "Saved " + saved + " of " + batch.length + ". Not saved: " + failed.join("; ") +
+          ". Nothing you entered was cleared; reload to see what was saved."
+        : "Saved " + saved + " of " + batch.length + ". You changed an answer while " +
+          "that was saving, and it is not saved yet: press Submit again to save it.";
+      st.hidden = false;
+    });
+  });
+
   function start() {
     var name = boothName();
     if (!name || !document.body) return;
diff --git a/booth/templates/_ask_inline.html b/booth/templates/_ask_inline.html
index 08bbd45..6e41b85 100644
--- a/booth/templates/_ask_inline.html
+++ b/booth/templates/_ask_inline.html
@@ -62,6 +62,9 @@
               placeholder="{{ a.notes_label }} (optional)">{{ a.answer.notes if a.answer else '' }}</textarea>
   {% endif %}
   <button type="submit" class="bk-ask-go" form="{{ form_id }}">{% if a.answer %}Update answer{% else %}Submit answer{% endif %}</button>
+  {# Empty and hidden: where embed.js says which asks a several-at-once submit
+     could not save. The script only sets its text; it builds no markup. #}
+  <p class="bk-ask-status" role="status" hidden></p>
 </div>
 {% endmacro %}
 
diff --git a/booth/templates/base.html b/booth/templates/base.html
index 1a34f85..154efd6 100644
--- a/booth/templates/base.html
+++ b/booth/templates/base.html
@@ -1064,6 +1064,12 @@
       var st = document.querySelector('[data-region="status"]');
       if (st) { st.textContent = text; st.hidden = false; }
     }
+    /* A new save clears the last one's words: a partial batch's "not saved"
+       must not outlive the save that fixes it. */
+    function quiet() {
+      var st = document.querySelector('[data-region="status"]');
+      if (st) { st.textContent = ''; st.hidden = true; }
+    }
     /* A form's IDENTITY: its action plus the hidden fields that say what it is
        about (which pick, which item, which mark). Stable across renders, where
        a position inside a region is not — a form that appears or vanishes
@@ -1103,9 +1109,12 @@
        must not (a revealed blur, a closed doc, a disclosure the reader
        opened or closed); and every DIRTY control — a half-typed note, an
        edited one, a radio picked and not yet sent. All matched by IDENTITY,
-       never by position. The form just sent is the exception: its fields and
-       its disclosure come back as the server rendered them. */
-    function carry(oldEl, newEl, sent) {
+       never by position. The forms just sent are the exception: their fields
+       and their disclosures come back as the server rendered them — unless the
+       form changed after its press, when it carries like any unsent form
+       (heid bug-hunt, hulda: a pick changed mid-flight came back as the saved
+       copy of the earlier one). See isSent. */
+    function carry(oldEl, newEl, isSent) {
       var olds = [].slice.call(oldEl.querySelectorAll('img[src], video[src], audio[src]'));
       newEl.querySelectorAll('img[src], video[src], audio[src]').forEach(function (m) {
         for (var i = 0; i < olds.length; i++) {
@@ -1123,7 +1132,7 @@
       var oldDetails = detailsMap(oldEl);
       Object.keys(oldDetails).forEach(function (k) {
         var d = oldDetails[k];
-        if (sent && d.contains(sent)) return;
+        if ([].some.call(d.querySelectorAll('form'), isSent)) return;
         if (freshDetails[k]) freshDetails[k].open = d.open;
       });
       var freshFields = {};
@@ -1131,7 +1140,7 @@
         if (el.type !== 'hidden') freshFields[fieldKey(el)] = el;
       });
       oldEl.querySelectorAll('textarea, input').forEach(function (el) {
-        if (el.type === 'hidden' || (sent && el.form === sent)) return;
+        if (el.type === 'hidden' || isSent(el.form)) return;
         var t = freshFields[fieldKey(el)];
         if (!t) return;
         if (el.type === 'radio' || el.type === 'checkbox') {
@@ -1145,7 +1154,7 @@
        beyond a tile falling out of a filter — a panel region that appeared or
        vanished — or when the page has no region to swap at all. Then only a
        reload tells the truth. */
-    function swap(html, sent) {
+    function swap(html, isSent) {
       var fresh = new DOMParser().parseFromString(html, 'text/html');
       var freshById = {}, liveIds = {};
       fresh.querySelectorAll('[data-region]').forEach(function (c) {
@@ -1164,7 +1173,7 @@
         var next = freshById[id];
         if (next) {
           var node = document.importNode(next, true);
-          carry(el, node, sent);
+          carry(el, node, isSent);
           el.replaceWith(node);
           swapped++;
         } else if (id.indexOf('item-') === 0) {
@@ -1184,37 +1193,162 @@
     /* SERIALIZED: each save runs its POST, its re-fetch and its swap before
        the next begins, so an older snapshot can never land after a newer one.
        A form already queued or in flight ignores another submit — a
-       double-click writes one note, not two. */
+       double-click writes one note, not two. IN FLIGHT is keyed by the form's
+       IDENTITY (formKey), never marked on the node: a queued save's swap
+       replaces the node with a fresh copy, and a mark on the old node would
+       let a second press send the same answer again (heid bug-hunt, hulda). */
     var queue = Promise.resolve();
-    function run(form, data) {
+    var pending = {};
+    /* The one place "which form is in flight" gets its identity. */
+    function flightKey(f) { return formKey(f); }
+    function post(form, data) {
       return fetch(form.action, {
         method: 'POST', body: new URLSearchParams(data),
         headers: {'Accept': 'application/json'}, credentials: 'same-origin'
       }).then(function (r) {
         if (r.status !== 204) throw new Error('status ' + r.status);
-        return fetch(window.location.href, {headers: {'Accept': 'text/html'}, credentials: 'same-origin'});
-      }).then(function (r) {
-        if (!r.ok) throw new Error('status ' + r.status);
-        return r.text();
-      }).then(function (html) {
-        if (!swap(html, form)) window.location.reload();
-      }).catch(function () {
-        /* Said, then reloaded after a beat, so the words are readable rather
-           than a flash before the page goes. */
-        say('Could not save in place — reloading to show what was saved.');
-        setTimeout(function () { window.location.reload(); }, 900);
+      });
+    }
+    function serial(f) {
+      return new URLSearchParams(new FormData(f)).toString();
+    }
+    /* What was sent, as the swap needs it: each form by IDENTITY (flightKey),
+       with its fields as they stood at the press. A live form is "just sent"
+       only if it is one of those AND unchanged since — so a queued save whose
+       node an earlier swap replaced is still recognised (hulda: its saved
+       note came back as a draft), and a form edited mid-flight is not. */
+    function sentSet(recs) {
+      return function (f) {
+        if (!f) return false;
+        var k = flightKey(f);
+        for (var i = 0; i < recs.length; i++) {
+          if (recs[i].key === k && recs[i].snap === serial(f)) return true;
+        }
+        return false;
+      };
+    }
+    /* Resolves true when the fresh page was placed. `keep` (a batch): never
+       reload — a structural change or a failed GET is said, not acted on,
+       because a reload would take every unsent draft with it. */
+    function refresh(recs, keep) {
+      return fetch(window.location.href, {headers: {'Accept': 'text/html'}, credentials: 'same-origin'})
+        .then(function (r) {
+          if (!r.ok) throw new Error('status ' + r.status);
+          return r.text();
+        }).then(function (html) {
+          if (swap(html, sentSet(recs))) return true;
+          if (!keep) window.location.reload();
+          return false;
+        });
+    }
+    function fail() {
+      /* Said, then reloaded after a beat, so the words are readable rather
+         than a flash before the page goes. */
+      say('Could not save in place — reloading to show what was saved.');
+      setTimeout(function () { window.location.reload(); }, 900);
+    }
+    function run(form, data, snap) {
+      return post(form, data).then(function () {
+        return refresh([{key: flightKey(form), snap: snap}]);
+      }).catch(fail);
+    }
+    /* SEVERAL PICKS AT ONCE (C3 step 3a, 2026-09-27). One pick is one form
+       is one POST, so a page holding several picks saved only the one whose
+       button was pressed. The others' unsent radios SURVIVED the swap (carry,
+       above) but were never saved — and pressing a BLANK pick's button was a
+       400, whose failure reload wiped them all. The operator's report: one
+       submit must save every answer he filled in.
+
+       A pick form carries a hidden `ask`, the field formKey reads. DIRTY is
+       what carry() measures: a control that differs from its server-rendered
+       default. */
+    function isPick(f) {
+      return !!f.querySelector('input[type=hidden][name="ask"]');
+    }
+    function dirty(f) {
+      return [].some.call(f.elements, function (el) {
+        if (el.type === 'radio' || el.type === 'checkbox') return el.checked !== el.defaultChecked;
+        if (el.tagName === 'TEXTAREA' || el.type === 'text') return el.value !== el.defaultValue;
+        return false;
+      });
+    }
+    /* What a submit on pick form `form` must send: every dirty pick form of
+       the same action NOT already in flight, in DOCUMENT ORDER — `form` only
+       if it is dirty, since re-sending an untouched answer re-dates it and a
+       blank one is refused. Null when no OTHER pick form is dirty, which
+       leaves the one-form path below exactly as it was. A form in flight
+       still COUNTS as another dirty form, so a press on a clean pick during a
+       batch is an empty batch — a no-op — and never the blank one-form POST
+       whose 400 would reload the page mid-save (heid bug-hunt, kimi). */
+    function pickBatch(form) {
+      if (!isPick(form)) return null;
+      var action = form.getAttribute('action');
+      var picks = [].filter.call(document.querySelectorAll('form[data-inplace]'), function (f) {
+        return f.getAttribute('action') === action && isPick(f);
+      });
+      var others = picks.some(function (f) { return f !== form && dirty(f); });
+      return others ? picks.filter(function (f) { return dirty(f) && !pending[flightKey(f)]; }) : null;
+    }
+    function askOf(f) {
+      var h = f.querySelector('input[type=hidden][name="ask"]');
+      return h ? h.value : '?';
+    }
+    /* One POST per form, in turn, a refusal stopping none of the rest (one
+       stale pick must not cost the others). Then ONE refresh in place, in
+       which only the forms the server TOOK count as sent, so everything else
+       carries by identity — a refused pick's input and any draft included.
+       A batch NEVER reloads the page (heid bug-hunt, 3 of 4 arms, then
+       hulda): a refusal, a failed refresh or a changed page is SAID, in the
+       status line, and the page stays. Never a re-POST. */
+    function runAll(forms, datas, snaps) {
+      var saved = [], refused = [], chain = Promise.resolve();
+      forms.forEach(function (f, i) {
+        chain = chain.then(function () {
+          return post(f, datas[i]).then(function () {
+            saved.push({key: flightKey(f), snap: snaps[i]});
+          }, function (e) { refused.push(askOf(f) + ' (' + e.message + ')'); });
+        });
+      });
+      return chain.then(function () {
+        var told = refused.length
+          ? 'Saved ' + saved.length + ' of ' + forms.length + '. Not saved: ' +
+            refused.join('; ') + '. Nothing else you entered was cleared.'
+          : '';
+        var shown = saved.length ? refresh(saved, true) : Promise.resolve(true);
+        return shown.then(function (placed) {
+          if (!placed) say((told || 'Saved.') + ' The page changed meanwhile; reload to see it.');
+          else if (told) say(told);
+        }, function () {
+          say((told || 'Saved ' + saved.length + ' of ' + forms.length + '.') +
+              ' Could not refresh the page; reload to see what was saved.');
+        });
       });
     }
     document.addEventListener('submit', function (ev) {
       var form = ev.target;
       if (!form.matches || !form.matches('form[data-inplace]') || ev.defaultPrevented) return;
       ev.preventDefault();
-      if (form.__busy) return;
-      form.__busy = true;
+      if (pending[flightKey(form)]) return;
+      var batch = pickBatch(form);
+      if (batch) {
+        if (!batch.length) return;          /* everything dirty is already in flight */
+        quiet();
+        var keys = batch.map(flightKey);
+        var datas = batch.map(function (f) { return new FormData(f); });   /* read NOW */
+        var snaps = batch.map(serial);
+        keys.forEach(function (k) { pending[k] = true; });
+        queue = queue.then(function () { return runAll(batch, datas, snaps); })
+                     .then(function () { keys.forEach(function (k) { delete pending[k]; }); });
+        return;
+      }
+      var key = flightKey(form);
+      pending[key] = true;
+      quiet();
+      var snap = serial(form);             /* the form's own fields, before the submitter */
       var data = new FormData(form);
       if (ev.submitter && ev.submitter.name) data.append(ev.submitter.name, ev.submitter.value);
-      queue = queue.then(function () { return run(form, data); })
-                   .then(function () { form.__busy = false; });
+      queue = queue.then(function () { return run(form, data, snap); })
+                   .then(function () { delete pending[key]; });
     });
   })();
 </script>
diff --git a/docs/contracts/r2_flow.contract.md b/docs/contracts/r2_flow.contract.md
index c5c0305..593d600 100644
--- a/docs/contracts/r2_flow.contract.md
+++ b/docs/contracts/r2_flow.contract.md
@@ -215,11 +215,77 @@ today's zoom flag form carries no `back`, so it lands on the gallery.
      by the pick or form it holds. A flag that adds a tray row above a draft
      must not move the draft into the wrong box. The form just sent is the
      exception: its fields come back as the server rendered them, and its
-     disclosure comes back folded.
+     disclosure comes back folded — UNLESS it changed after the press, when it
+     carries like any unsent form (amended 2026-09-27; a pick changed
+     mid-flight came back as the saved copy of the earlier one). "Just sent"
+     is the form's IDENTITY plus its fields as they stood at the press, never
+     the DOM node: a queued save whose node an earlier swap replaced is still
+     recognised, where a node test missed it and carried a saved note's text
+     back as a draft.
 3. **Saves are SERIALIZED.** Each save runs its POST, its GET and its swap
    before the next begins, so an older snapshot never lands after a newer one
    (three quick flags show three flags). A form already queued or in flight
-   ignores another submit: a double-click writes one note, not two.
+   ignores another submit: a double-click writes one note, not two. "In flight"
+   is keyed by the form's IDENTITY (the same action + hidden-field key the
+   carry uses), never marked on the DOM node, because a queued save's swap
+   replaces the node with a fresh copy (amended 2026-09-27).
+3a. **Several picks at once (amended 2026-09-27).** A pick form is a
+   `data-inplace` form carrying a hidden `ask` field. It is **dirty** when any
+   control in it differs from its server-rendered default (`checked` vs
+   `defaultChecked`, `value` vs `defaultValue`). A submit on a pick form while
+   ANOTHER pick form of the same action is dirty sends every dirty pick form
+   NOT already in flight — the pressed one only if it is dirty.
+   - A form in flight still counts as "another dirty form", so a press on a
+     clean pick during a batch is an empty batch: a no-op, never the blank
+     one-form POST whose 400 would take step 4 mid-save.
+   - One POST per form, to its own action, with its own fields read at the
+     moment of the press, one after another in DOCUMENT ORDER of the forms. A
+     refused POST does not stop the ones after it.
+   - None refused: ONE GET and ONE swap, in which every form sent counts as
+     "the form just sent": its fields come back as the server rendered them,
+     its disclosure folded.
+   - Any refused: ONE GET and ONE swap in which only the forms the server
+     TOOK count as sent, so everything else carries by identity — the refused
+     pick's own input and any draft on the page included — then the status
+     line says how many saved and names each pick that did not, with the
+     reason. (Nothing saved at all: no GET, just the words.) A pick withdrawn
+     under the page has no form in the fresh page to carry into; the reason
+     says so. This is the same rule as the verbatim half: a refusal never
+     clears what the operator entered.
+   - **A batch never reloads.** Where step 2 would reload — a failed GET, or a
+     fresh page whose structure changed — a batch says so in the status line
+     ("reload to see it") and keeps the page, because a reload would take
+     every unsent draft with it. Step 4's say-and-reload stays the one-form
+     path's alone.
+   - The status line is cleared when the next save starts, so a "not saved"
+     never outlives the save that fixes it.
+   - With no other dirty pick form, the submit takes steps 1–3 exactly as
+     before.
+
+   Why: C3 already CARRIED an unsent pick across another save, so it survived
+   — but it was never SAVED, and pressing the submit of a BLANK pick got a
+   400, whose failure reload wiped every one. The operator's report
+   (2026-09-27, relayed by infra-ops): one submit on a page must save every
+   answer he filled in. The verbatim half of the same fix is U3's "Submitting
+   several asks at once"; the two surfaces share the dirty rule, the order and
+   the refusal rule, and differ only where their machinery does (this one
+   swaps in place; the verbatim page reloads when nothing is left unsaved).
+   *Falsifiable* (`tests/mutations/r2_submit_all.toml`): ignore the other pick
+   forms and `test_one_submit_on_the_marks_page_saves_every_changed_pick`
+   fails; send the pressed form even when blank and
+   `test_pressing_a_blank_picks_submit_saves_the_others_and_skips_it` fails;
+   stop at the first refusal, reload on one, or count a refused pick as sent,
+   and `test_a_refused_pick_in_a_batch_costs_only_itself_and_clears_nothing`
+   fails; stop counting a form in flight as dirty and
+   `test_pressing_a_clean_pick_during_a_batch_sends_nothing` fails; key "in
+   flight" on the DOM node and
+   `test_a_pick_in_flight_stays_in_flight_across_another_saves_swap` fails;
+   leave the status line up and `test_a_later_save_clears_a_stale_not_saved_line`
+   fails; reload when the refresh fails and
+   `test_a_batch_whose_refresh_fails_keeps_the_page` fails; count a sent form
+   as sent after it changed and
+   `test_a_change_to_a_sent_pick_during_the_flight_is_kept` fails; count no
+   form as sent and `test_a_saved_notes_box_comes_back_empty` fails.
 4. **The script never re-POSTs.** A retry after a lost response would re-apply
    the judgment: a duplicate note, or a re-dated answer.
    - On a non-204 HTTP response, or a network failure, it writes a fixed
diff --git a/docs/contracts/u3_declared_embed_seam.contract.md b/docs/contracts/u3_declared_embed_seam.contract.md
index 44f49cb..8d0e7c9 100644
--- a/docs/contracts/u3_declared_embed_seam.contract.md
+++ b/docs/contracts/u3_declared_embed_seam.contract.md
@@ -235,6 +235,86 @@ It is three lines, it costs nothing, and the sensitivity floor of that probe is
 guards against is a form the operator fills in whose controls reach no form,
 so the button does nothing.
 
+## Submitting several asks at once (amended 2026-09-27)
+
+**The defect.** One pick is one `<form>` is one POST to `/answer`, and that
+POST 303s back to the page. On a report carrying several picks, a submit sent
+exactly ONE of them, and the reload that followed wiped every pick the operator
+had made in the others. His report, relayed by infra-ops: *"I go through, submit
+a question and it only submits the last one and clears out the top ones."*
+Confirmed against the live `auk-audition` booth (three single-question picks,
+no anchors, so all three in the tail) before any code: the access log shows one
+POST at 15:02:23 that saved the LAST pick on the page, the reload, then a 400
+four seconds later — the submit of a pick the reload had just blanked — and the
+other two re-answered one at a time. **The server is not the defect**: every
+POST did exactly what `/answer` promises. The page gave him one button per
+form and no way to send them together.
+
+**The rule.** embed.js listens for `submit` on the forms IT mounted (never an
+author's form). A form is **dirty** when any control it owns — `form.elements`,
+which includes every control bound to it by `form=` wherever it sits — differs
+from its server-rendered default: a radio or checkbox whose `checked` differs
+from `defaultChecked`, a textarea or text input whose `value` differs from
+`defaultValue`.
+
+```
+submit on one of our forms F:
+  a batch is in flight            -> preventDefault; nothing else (never the
+                                     browser's POST racing the batch)
+  no OTHER of our forms is dirty  -> do nothing; the browser's own POST and 303,
+                                     exactly as before this amendment
+  otherwise                       -> preventDefault, and send EVERY dirty form of
+                                     ours, F included only if F is dirty
+  send: one POST per form, to that form's own action, carrying that form's own
+        FormData read AT THE PRESS, with `Accept: application/json` (the
+        route's 204, r2 C3), one after another, in DOCUMENT ORDER of the <form>
+        elements. A refused form does not stop the ones after it. A form the
+        server took (204) gets a new baseline: what it sent. From then on it is
+        dirty only if it differs from THAT.
+  then: no refusal AND nothing -> reload the page (a GET), so what shows is
+        of ours is dirty           the server's record
+        otherwise                -> NO reload. The pressed form's submit block
+          (a refusal, or a change    says how many saved and what did not, with
+          made during the flight)    the server's reason, and everything the
+                                     operator entered stays on the page.
+        A refusal blocks the reload ON ITS OWN: a refused form set back to its
+        first value reads clean, and "nothing dirty" alone reloaded over it.
+```
+
+Five consequences, each deliberate:
+
+- **A blank pick is skipped, never refused.** A pick with nothing entered is not
+  dirty and is not sent, so pressing its button no longer produces the 400 page
+  the log shows. Blanks stay legal, as `build_answer` has held since 2026-09-09.
+- **A pick nobody touched is not re-sent — including the one whose button was
+  pressed, and including one this page already saved.** Re-sending an answer
+  re-dates it, and a reading session would see a fresh answer that nobody gave.
+  The moved baseline is what keeps a retry after a partial refusal to exactly
+  the picks that did not save; it also means correcting a saved pick back to
+  its first-rendered value counts as a change and is sent.
+- **One refused pick costs only itself.** A pick withdrawn or re-declared while
+  the page was open is refused (404 / 400); the rest are saved regardless. This
+  is the partial-answer ruling's reasoning applied one level up: refusing
+  everything because one was stale throws away the ones that were made.
+- **The page is reloaded only when nothing would be lost by it.** The page
+  stays live while the batch is in flight; a pick made or a note typed in that
+  window is unsaved input, and a reload would clear it — the exact loss this
+  amendment exists to stop. So the reload waits on "every POST succeeded" AND
+  "nothing of ours is dirty" — each on its own. The saved picks' tags stay stale until he
+  reloads, and the message says so. Nothing is ever re-sent without a fresh
+  press; a press after a lost response may re-send (and re-date) a pick that
+  did land, which is the price of never retrying on our own.
+- **A press during the flight is ignored.** Checked before anything else, so a
+  press on a form with no other dirty form beside it cannot fall through to the
+  browser's POST while the batch runs.
+
+**What it does not change.** `/answer`, its fields, its 204 and its 303 are
+untouched: the server has no batch endpoint and no new request shape. A page
+whose only dirty form is the pressed one gets the plain form submission,
+byte-identical to before. The status line is server-rendered, empty and
+`hidden` in the `submit` macro; the script only sets its text, so embed.js
+still renders no markup of an ask.
+
 **Step 5 deletes an element.** Today `inject_asks` injects `<a id="bk-ask-<id>-top">`
 before the first fragment of each pick so the chip has somewhere to jump. The
 fragments already carry ids; document order in a live DOM is directly queryable;
@@ -336,6 +416,32 @@ rather than strict.
 and `test_partially_marked_page_still_shows_every_question` fails in the browser
 with 2 of 4 radio groups present.
 
+**INV-8 — One submit saves every pick on the page the operator changed
+(amended 2026-09-27).** Per "Submitting several asks at once": every dirty form
+of ours is sent, in document order; a clean one never is, nor a saved one
+again; a refused one stops nothing; the page reloads only when nothing was
+refused and nothing of ours is left unsaved; a press during the flight is ignored; and with no other dirty
+form the submit is the browser's own.
+*Falsifiable*, one change per clause, each in `tests/mutations/u3_submit_all.toml`:
+send only the pressed form and
+`test_one_submit_saves_every_answered_ask_on_the_page` fails; send the pressed
+form whether or not it is dirty and `test_a_blank_ask_is_skipped_never_refused`
+fails; send every form regardless and `test_an_ask_nobody_touched_is_not_re_sent`
+fails; intercept a lone dirty form and `test_one_changed_ask_still_submits_as_a_plain_form`
+fails; send in reverse and `test_the_asks_are_sent_in_document_order` fails;
+stop at the first refusal, reload on one, or never show the status line, and
+`test_a_refused_ask_costs_only_itself_and_clears_nothing` fails; listen to every
+form on the page, or trust our id prefix without the mounted-root check, and
+`test_an_authors_own_form_is_never_taken_over` fails; let a press in flight fall
+through and `test_a_press_inside_the_flight_never_fires_a_native_post` fails;
+reload when every POST succeeded regardless of what changed meanwhile and
+`test_input_made_during_the_flight_is_kept_and_saved_on_the_next_press` fails;
+leave a saved form's baseline where it was and
+`test_a_retry_after_a_refusal_sends_only_what_did_not_save` fails; drop the
+class-level `[hidden]` rule and `test_the_empty_status_line_stays_hidden_under_host_css`
+fails; let "nothing dirty" alone decide the reload and
+`test_a_refusal_blocks_the_reload_even_when_nothing_reads_dirty` fails.
+
 ## Out of scope (deferred or never)
 
 Named so a reviewer does not read them as drift.
diff --git a/persistent-memory.d/2026-09-27-one-submit-saves-every-ask.md b/persistent-memory.d/2026-09-27-one-submit-saves-every-ask.md
new file mode 100644
index 0000000..a25f50a
--- /dev/null
+++ b/persistent-memory.d/2026-09-27-one-submit-saves-every-ask.md
@@ -0,0 +1,102 @@
+# 2026-09-27 — One submit saves every ask on the page
+
+**The report.** Prime to infra-ops, relayed to booth-dev (althing thread
+`01M3JED397G1SZH7580PCXNVVA`): "submitting a question should go through and
+submit ALL answers. As it is, I go through, submit a question and it only
+submits the last one and clears out the top ones."
+
+**Confirmed against live data before any code, and it matched to the second.**
+`auk-audition`: three single-question picks, no anchors, so all three in the
+embed tail in `(created, id)` order. The access log: one POST at 15:02:23 (303)
+that saved `auk-emotion`, the LAST pick on the page; the reload; a POST at
+15:02:27 that 400'd (the submit of a pick the reload had just blanked — the
+browser showed a raw `{"detail": ...}` page); then `auk-clone` at 15:04:45 and
+`auk-events` at 15:05:02, answered one at a time. infra-ops' reading of the
+code was right. `vastblue-site-visit-2026-09-29` carries 14 picks and would
+have hit it next.
+
+**The cause is not the server.** One pick = one `<form>` = one POST to
+`/answer`, and every POST did what `/answer` promises. The page offered one
+button per form and no way to send them together. Two surfaces, two machineries:
+
+- **Verbatim report (embed.js).** Plain form POST + 303 reload. The reload
+  wiped every unsent pick.
+- **Booth pages (base.html's in-place script: marks page, lightbox verdict
+  aside, review rail).** R2 C3 already CARRIED unsent picks across a swap, so
+  they survived — but were never SAVED. And pressing a blank pick's submit was
+  a 400, whose failure path reloads and wipes them all.
+
+**The shape (booth-dev's call; infra-ops said "the shape is your call").**
+Client-side, both surfaces, no server change: a submit on a pick form, while
+ANOTHER pick form on the page is dirty (a control differs from its
+server-rendered default), sends every dirty pick form — the pressed one only if
+dirty — one POST each to the unchanged `/answer` with `Accept:
+application/json` (the 204), serially in document order, a refusal stopping
+none of the rest. With no other dirty form, nothing changes: the browser's own
+POST (verbatim) or C3's one-form path (Booth pages).
+
+- **Rejected: a server batch endpoint + one page-level form.** It would also
+  fix no-JS, but needs ask-scoped field names, a single `<form>` element placed
+  by embed.js, and rewires `formKey` identity on the Booth pages (one form
+  holding many `ask` fields). Large blast radius for a no-JS verbatim path that
+  does not exist (U3's named cost). Atomic all-or-nothing was also the WRONG
+  semantics: one stale pick would refuse the rest — the 2026-09-09
+  partial-answer ruling's reasoning, one level up.
+- **Untouched picks are not re-sent**, the pressed one included: re-sending
+  re-dates an answer nobody gave.
+- **A refusal never clears what was entered, on either surface** (the first
+  cut had the Booth batch take C3's say-and-reload path; heid's panel caught
+  it, see below). Verbatim: no reload while anything of ours is dirty; the
+  pressed form's server-rendered, empty, hidden `.bk-ask-status` line says what
+  did not save. Booth pages: refresh in place with only the forms the server
+  took counted as sent, so the refused pick and any draft carry by identity;
+  the status region names what did not save. C3's ONE-form failure path
+  (step 4, say and reload) is untouched — not this change's surface, and it
+  still loses drafts on a refusal (named, not fixed).
+
+**The heid bug-hunt panel (4/4 arms, thread `01M3JFDM1AWT2TCKS6E9NJM9G4`) was
+the gate that paid.** All four landed on the same blind spot: the batch reads
+every form AT THE PRESS but the page stays live through the flight, and every
+guard protecting that window (`sending`, `__busy`, `held`) SURVIVED mutation,
+because no test pressed or edited inside a flight. Six of its rows reproduced
+RED in a browser before any fix: a successful embed batch reloading away a pick
+made mid-flight (S1); the Booth batch's refusal reload (S2); saved forms staying
+dirty so a retry re-dated them (S3); in-flight marked on DOM nodes a queued
+swap replaced (S5 — now keyed by `formKey` identity via `flightKey`); a press
+in flight falling through to a native POST (embed) or a blank 400 (Booth) (S6).
+The trick that made them testable: hold every POST's reply 700ms in the CLIENT
+(`_HOLD_POSTS`, the `_HOLD_FIRST_REFRESH` pattern) so the window is wide enough
+to act in on purpose. S7 (no ask dedupe) rejected as unreachable; S8–S11
+accepted with reasons in the fold reply (`01M3JHYKA0GSJG4F836J1Z6TJS`).
+**Lesson: a feature that opens an async window needs a test that acts inside
+it; a green suite of single presses says nothing about the window.**
+
+**Round two: a single cold Hulda arm on the FOLDED tree found six more**
+(thread `01M3JHYKCPXFA34XMRCEW4P3DJ`), all in how the fold's own rules
+interacted — and Heid's primed second voice, reading only the fold's delta,
+cleared two of them wrongly and retracted. Four reproduced RED: a sent pick
+changed mid-flight came back as the saved copy (#1); a note queued behind a
+flag carried back as a draft because "just sent" was a DOM-node test (#3);
+a batch whose page GET failed reloaded drafts away (#4); the embed reloaded
+over a refusal the operator had set back to its first value (#6). Fix:
+"just sent" = `flightKey` identity + the form's serialization at the press
+(`sentSet`), a batch never reloads, and a refusal blocks the embed reload on
+its own. #2 was a message fix (a withdrawn pick's input has nowhere to go);
+#5 accepted (needs a refused POST AND a failed GET; costs a re-date of
+identical content). **A cold read of the whole diff beat a primed read of the
+delta** — worth remembering before scoping a re-review to "just the fold".
+
+**Contracts amended in the same commit:** U3 "Submitting several asks at once"
++ INV-8; R2 C3 steps 3 and 3a. **Mutation tables:** `tests/mutations/u3_submit_all.toml`
+(15 rows) and `tests/mutations/r2_submit_all.toml` (11 rows), all proved; the
+r2_flow row anchored on `form.__busy` moved to `pending[flightKey(form)]`.
+**Tests:** 23 new browser tests plus two tightened (the author-form test now
+wears our id prefix; the one-form failure test now asserts the reload it names).
+Suite 928 → 951.
+The `[a3]` case reproduced the live log exactly (only a3 saved).
+
+**Not done, named:** no no-JS batch on the Booth pages (each plain form still
+saves one pick with scripts off — the no-JS path is unchanged, as asked); a
+verbatim batch that keeps the page leaves the saved picks' tags stale until a
+reload, and the message says so; C3's one-form refusal still reloads drafts
+away; Hulda #5 (above) accepted.
diff --git a/persistent-memory.md b/persistent-memory.md
index 26937d8..e0184a6 100644
--- a/persistent-memory.md
+++ b/persistent-memory.md
@@ -1,6 +1,6 @@
 # Persistent memory — booth
 
-_Last updated: 2026-09-25_
+_Last updated: 2026-09-27_
 
 > **Always check for `/tmp/booth-dev-handoff.md`** — if it exists and its
 > `Written:` stamp is under 8 hours old, read it (it carries the in-flight
@@ -17,12 +17,14 @@ loop it turned out to actually be.
 
 ## Current state / in-flight
 
-_As of 2026-09-25:_
+_As of 2026-09-27:_
 
-- 🟢 **NOTHING IS IN FLIGHT** (shutdown snapshot, 2026-09-25). The tree is
-  clean, no branch or worktree is open, and no peer is waiting on booth-dev.
-  main is ONE memory commit ahead of origin (this snapshot), unpushed: the
-  push is the operator's call.
+- ✅ **ONE SUBMIT SAVES EVERY ASK ON A PAGE** (2026-09-27, Prime's bug via
+  infra-ops, thread `01M3JED397G1SZH7580PCXNVVA`). Client-side on both
+  surfaces (embed.js, base.html's in-place script), no server change; a
+  refusal or a mid-flight edit never clears input. Local commits on main,
+  UNPUSHED (push is the operator's call); service restarted to make it live.
+  → `persistent-memory.d/2026-09-27-one-submit-saves-every-ask.md`
 - ✅ **r3 COMPARE IS LIVE AND PUSHED** (operator: "merge r3 once the mutation
   check is clean", then "merge and push", 2026-09-24). origin/main is the r3
   arc's tip: r3 (`f8d136a`), design-dev's race fix (`d54bb04`: one compare ring
@@ -85,6 +87,7 @@ _As of 2026-09-25:_
 
 ## Recent decisions
 
+- `[2026-09-27]` ✅ **One submit saves every ask on the page; the heid panel found the async window** — READ BEFORE TOUCHING THE SUBMIT PATH OF embed.js OR base.html: a batch reads forms at the press while the page stays live, and only a test that acts inside the flight can see it → `persistent-memory.d/2026-09-27-one-submit-saves-every-ask.md`
 - `[2026-09-24]` ⏸ **The upload route's three lifecycle gaps: DEFERRED** — the pickup-id `mkdir` sits outside the try (a FileExistsError race), `rmtree(ignore_errors=True)` hides its own failure, and `except Exception` misses CancelledError. All three are rare; the operator was told and merged without them. Tracked in `225ba32`'s commit message.
 - `[2026-09-24]` ✅ **Upload names: two crashes found, then two holes in the fix** — READ BEFORE WRITING A SANITISER: drop everything droppable FIRST, then apply the structural rules; a NUL test through httpx `files=` proves nothing → `persistent-memory.d/2026-09-24-upload-names-two-crashes-then-two-holes.md`
 - `[2026-09-24]` ✅ **The r3 seam pass: what only it could see** — 12 contract-vs-code mismatches folded before code. READ BEFORE A CONTRACT THAT ADDS `data-region`S, MOVES TEMPLATE CODE, OR ADDS A `/b/{name}/<word>` ROUTE → `persistent-memory.d/2026-09-24-r3-seam-pass-what-only-it-could-see.md`
diff --git a/tests/mutations/r2_flow.toml b/tests/mutations/r2_flow.toml
index d6fc463..d71845b 100644
--- a/tests/mutations/r2_flow.toml
+++ b/tests/mutations/r2_flow.toml
@@ -118,7 +118,7 @@ new = '''        except OSError:
 label = 'C3 client: no busy guard (a double-click writes twice)'
 file = "booth/templates/base.html"
 test = "tests/test_flow_browser.py::test_an_unsaved_choice_survives_a_save_elsewhere_and_a_double_click_writes_once"
-old = '''      if (form.__busy) return;
+old = '''      if (pending[flightKey(form)]) return;
 '''
 new = ''''''
 
@@ -133,8 +133,8 @@ new = ''''''
 label = 'C3 client: saves are not serialized'
 file = "booth/templates/base.html"
 test = "tests/test_flow_browser.py::test_quick_successive_flags_all_show"
-old = '''      queue = queue.then(function () { return run(form, data); })'''
-new = '''      queue = run(form, data)'''
+old = '''      queue = queue.then(function () { return run(form, data, snap); })'''
+new = '''      queue = run(form, data, snap)'''
 
 [[mutation]]
 label = 'C3 the standalone marks page has no region'
diff --git a/tests/mutations/r2_submit_all.toml b/tests/mutations/r2_submit_all.toml
new file mode 100644
index 0000000..a331e52
--- /dev/null
+++ b/tests/mutations/r2_submit_all.toml
@@ -0,0 +1,109 @@
+# R2 C3 step 3a, 2026-09-27: one submit saves every changed pick on a Booth
+# page (the marks page, the lightbox's verdict aside, the review rail). The
+# Booth-page half of the operator's report; the verbatim half is
+# u3_submit_all.toml. Contract: docs/contracts/r2_flow.contract.md, C3 step 3a.
+# The flight-window and refusal rows came from the heid bug-hunt panel on the
+# first cut.
+
+unit = "r2 submit all"
+
+[[mutation]]
+label = "the other pick forms are ignored (one form, one save, as before)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_one_submit_on_the_marks_page_saves_every_changed_pick"
+old = '''
+      var batch = pickBatch(form);'''
+new = '''
+      var batch = null;'''
+
+[[mutation]]
+label = "the pressed pick is sent even when blank (its 400 reloads the rest away)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_pressing_a_blank_picks_submit_saves_the_others_and_skips_it"
+old = '''
+      return others ? picks.filter(function (f) { return dirty(f) && !pending[flightKey(f)]; }) : null;'''
+new = '''
+      return others ? picks.filter(function (f) { return (f === form || dirty(f)) && !pending[flightKey(f)]; }) : null;'''
+
+[[mutation]]
+label = "a refusal stops the picks after it"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_refused_pick_in_a_batch_costs_only_itself_and_clears_nothing"
+old = '''
+          return post(f, datas[i]).then(function () {'''
+new = '''
+          if (refused.length) return;
+          return post(f, datas[i]).then(function () {'''
+
+[[mutation]]
+label = "a refused batch reloads (the refused pick and every draft are lost)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_refused_pick_in_a_batch_costs_only_itself_and_clears_nothing"
+old = '''
+        var shown = saved.length ? refresh(saved, true) : Promise.resolve(true);'''
+new = '''
+        if (refused.length) { fail(); return; }
+        var shown = saved.length ? refresh(saved, true) : Promise.resolve(true);'''
+
+[[mutation]]
+label = "a refused pick counts as sent, so its input comes back as the server has it"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_refused_pick_in_a_batch_costs_only_itself_and_clears_nothing"
+old = '''
+          }, function (e) { refused.push(askOf(f) + ' (' + e.message + ')'); });'''
+new = '''
+          }, function (e) { saved.push({key: flightKey(f), snap: snaps[i]}); refused.push(askOf(f) + ' (' + e.message + ')'); });'''
+
+[[mutation]]
+label = "a form in flight stops counting as another dirty form (a clean press 400s mid-save)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_pressing_a_clean_pick_during_a_batch_sends_nothing"
+old = '''
+        return f.getAttribute('action') === action && isPick(f);'''
+new = '''
+        return f.getAttribute('action') === action && isPick(f) && !pending[flightKey(f)];'''
+
+[[mutation]]
+label = "in flight is marked on the node, so a swap's fresh copy can be sent twice"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_pick_in_flight_stays_in_flight_across_another_saves_swap"
+old = '''
+    function flightKey(f) { return formKey(f); }'''
+new = '''
+    var nth = 0;
+    function flightKey(f) { return f.__fk || (f.__fk = 'n' + (++nth)); }'''
+
+[[mutation]]
+label = "a new save does not clear the last one's words"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_later_save_clears_a_stale_not_saved_line"
+old = '''
+      if (st) { st.textContent = ''; st.hidden = true; }'''
+new = ''''''
+
+[[mutation]]
+label = "a batch whose refresh fails reloads (every unsent draft with it)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_batch_whose_refresh_fails_keeps_the_page"
+old = '''
+        var shown = saved.length ? refresh(saved, true) : Promise.resolve(true);'''
+new = '''
+        var shown = saved.length ? refresh(saved, true).catch(function (e) { fail(); throw e; }) : Promise.resolve(true);'''
+
+[[mutation]]
+label = "a sent form counts as sent even when it changed after the press"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_change_to_a_sent_pick_during_the_flight_is_kept"
+old = '''
+          if (recs[i].key === k && recs[i].snap === serial(f)) return true;'''
+new = '''
+          if (recs[i].key === k) return true;'''
+
+[[mutation]]
+label = "no form counts as sent (a saved note's text carries back as a draft)"
+file = "booth/templates/base.html"
+test = "tests/test_flow_browser.py::test_a_saved_notes_box_comes_back_empty"
+old = '''
+          if (recs[i].key === k && recs[i].snap === serial(f)) return true;'''
+new = '''
+          if (false) return true;'''
diff --git a/tests/mutations/u3_submit_all.toml b/tests/mutations/u3_submit_all.toml
new file mode 100644
index 0000000..878a414
--- /dev/null
+++ b/tests/mutations/u3_submit_all.toml
@@ -0,0 +1,141 @@
+# U3 amendment, 2026-09-27: one submit saves every ask on a verbatim report.
+# The operator answered three asks top to bottom, pressed the last button, and
+# the 303 reload wiped the first two (`auk-audition`, 15:02:23). Contract:
+# docs/contracts/u3_declared_embed_seam.contract.md, "Submitting several asks at
+# once" and INV-8. Every row is a change tests/test_embed_browser.py claims to
+# forbid. The flight-window rows came from the heid bug-hunt panel on the first
+# cut, where every guard of that window survived its mutation.
+
+unit = "u3 submit all"
+
+[[mutation]]
+label = "only the pressed form is sent (the reported defect)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_one_submit_saves_every_answered_ask_on_the_page[a3]"
+old = '''
+    var batch = forms.filter(dirty);'''
+new = '''
+    var batch = [form];'''
+
+[[mutation]]
+label = "the pressed form is sent whether or not it is dirty (a blank one 400s)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_blank_ask_is_skipped_never_refused"
+old = '''
+    var batch = forms.filter(dirty);'''
+new = '''
+    var batch = forms.filter(function (f) { return f === form || dirty(f); });'''
+
+[[mutation]]
+label = "every form is sent, touched or not (re-dates an answer nobody gave)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_an_ask_nobody_touched_is_not_re_sent"
+old = '''
+    var batch = forms.filter(dirty);'''
+new = '''
+    var batch = forms;'''
+
+[[mutation]]
+label = "a lone dirty form is intercepted instead of left to the browser"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_one_changed_ask_still_submits_as_a_plain_form"
+old = '''
+    if (!others) return;                        // the browser's own POST and 303'''
+new = ''''''
+
+[[mutation]]
+label = "the forms are sent in reverse document order"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_the_asks_are_sent_in_document_order"
+old = '''
+    var batch = forms.filter(dirty);'''
+new = '''
+    var batch = forms.filter(dirty).reverse();'''
+
+[[mutation]]
+label = "a refusal stops the forms after it"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_refused_ask_costs_only_itself_and_clears_nothing"
+old = '''
+        return send(f, bodies[n]).then(function (why) {'''
+new = '''
+        if (failed.length) return;
+        return send(f, bodies[n]).then(function (why) {'''
+
+[[mutation]]
+label = "a refusal reloads the page and clears what was entered"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_refused_ask_costs_only_itself_and_clears_nothing"
+old = '''
+      if (!failed.length && !changed) { location.reload(); return; }'''
+new = '''
+      location.reload(); return;'''
+
+[[mutation]]
+label = "only dirtiness blocks the reload (a refused form set back to its first value reloads over the failure)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_refusal_blocks_the_reload_even_when_nothing_reads_dirty"
+old = '''
+      if (!failed.length && !changed) { location.reload(); return; }'''
+new = '''
+      if (!changed) { location.reload(); return; }'''
+
+[[mutation]]
+label = "a refusal is never said"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_refused_ask_costs_only_itself_and_clears_nothing"
+old = '''
+      st.hidden = false;'''
+new = ''''''
+
+[[mutation]]
+label = "an author's own form is taken over (no ownership check at the entry)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_an_authors_own_form_is_never_taken_over"
+old = '''
+    if (forms.indexOf(form) < 0) return;'''
+new = ''''''
+
+[[mutation]]
+label = "an author's form wearing our id prefix counts as ours (no mounted-root check)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_an_authors_own_form_is_never_taken_over"
+old = '''
+        if (ours[j].contains(all[i])) { out.push(all[i]); break; }'''
+new = '''
+        out.push(all[i]); break;'''
+
+[[mutation]]
+label = "a press during the flight falls through to the browser (a native POST races the batch)"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_press_inside_the_flight_never_fires_a_native_post"
+old = '''
+    if (sending) { ev.preventDefault(); return; }'''
+new = ''''''
+
+[[mutation]]
+label = "a change made during the flight is reloaded away"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_input_made_during_the_flight_is_kept_and_saved_on_the_next_press"
+old = '''
+      var changed = forms.some(dirty);'''
+new = '''
+      var changed = false;'''
+
+[[mutation]]
+label = "a saved form keeps its old baseline, so a retry re-sends (re-dates) it"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_a_retry_after_a_refusal_sends_only_what_did_not_save"
+old = '''
+          if (why === null) f.__bkSaved = bodies[n].toString();
+          else failed.push'''
+new = '''
+          if (why !== null) failed.push'''
+
+[[mutation]]
+label = "the empty status line shows under a host `p{display:block}`"
+file = "booth/static/embed.js"
+test = "tests/test_embed_browser.py::test_the_empty_status_line_stays_hidden_under_host_css"
+old = '''
+    ".bk-ask-status[hidden]{display:none}",'''
+new = ''''''
diff --git a/tests/test_embed_browser.py b/tests/test_embed_browser.py
index 2b18fa4..6ec6b09 100644
--- a/tests/test_embed_browser.py
+++ b/tests/test_embed_browser.py
@@ -670,3 +670,362 @@ def test_the_test_browser_has_no_internet(browser, live):
     page.close()
     assert "ERR_NAME_NOT_RESOLVED" in str(err.value) and external < 3, (str(err.value)[:80], external)
     assert local < 10, local
+
+
+# ---- one submit saves every ask on the page (2026-09-27) ---------------------
+#
+# The operator's report, relayed by infra-ops: "submitting a question should go
+# through and submit ALL answers. As it is, I go through, submit a question and
+# it only submits the last one and clears out the top ones." Confirmed against
+# the live `auk-audition` booth before any code: three single-question asks,
+# each its own <form>; the access log shows ONE POST at 15:02:23 saving the
+# LAST ask on the page, its 303 reload wiping the other two, then a 400 when a
+# now-blank ask was submitted. Contract: u3_declared_embed_seam, "Submitting
+# several asks at once".
+
+
+def _asks(booth, ids=("a1", "a2", "a3")):
+    """Single-question asks, declared in `ids` order, which is also their id
+    order, so `(created, id)` cannot disagree with the order written here."""
+    from booth.marks import declare_pick
+
+    booth.mkdir(parents=True, exist_ok=True)
+    for mid in ids:
+        declare_pick(booth, mid, {"prompt": f"About {mid}?", "options": ["yes", "no"]})
+    return booth
+
+
+def _answers(booth):
+    raw = json.loads((booth / ".marks.json").read_text())
+    return {m["id"]: m.get("answer") for m in raw["marks"]}
+
+
+def _choice(page, mid, value):
+    page.check(f'input[name="choice"][form="bk-ask-form-{mid}"][value="{value}"]')
+
+
+def _press(page, mid):
+    page.click(f"#bk-ask-{mid}-submit button.bk-ask-go")
+
+
+def _watch_posts(page):
+    """Every POST the page makes, as (resource type, ask id), in send order. A
+    native form submission is a `document` request; a script's is a `fetch`."""
+    from urllib.parse import parse_qs
+
+    posts = []
+
+    def seen(r):
+        if r.method == "POST":
+            ask = parse_qs(r.post_data or "").get("ask", [None])[0]
+            posts.append((r.resource_type, ask))
+    page.on("request", seen)
+    return posts
+
+
+PLAIN = f"<!doctype html><title>r</title><body><h1>R</h1>{SEAM}</body>"
+
+
+@pytest.mark.parametrize("pressed", ["a3", "a1"])
+def test_one_submit_saves_every_answered_ask_on_the_page(browser, live, pressed):
+    """The operator's exact sequence: answer top to bottom, press the LAST
+    submit. And the same from the FIRST button, because "press any one" is the
+    acceptance. Every ask he answered is recorded, and after the page comes
+    back every pick he made is still showing."""
+    base, data = live
+    b = _asks(data / "b")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    _choice(page, "a1", "yes")
+    _choice(page, "a2", "no")
+    _choice(page, "a3", "yes")
+    with page.expect_navigation():
+        _press(page, pressed)
+    got = _answers(b)
+    assert {k: (v or {}).get("choice") for k, v in got.items()} == \
+        {"a1": "yes", "a2": "no", "a3": "yes"}, got
+    page.wait_for_selector("#bk-ask-a3-submit")
+    showing = page.evaluate("""() => ['a1', 'a2', 'a3'].map(id => {
+        var c = document.querySelector('input[name="choice"][form="bk-ask-form-' + id + '"]:checked');
+        return c ? c.value : null; })""")
+    page.close()
+    assert showing == ["yes", "no", "yes"], f"a pick disappeared on reload: {showing}"
+
+
+def test_a_blank_ask_is_skipped_never_refused(browser, live):
+    """Pressing the submit of an ask he left blank used to be a 400 page
+    ("nothing to record"): the live log has one, four seconds after the reload
+    that wiped his picks. Blanks stay legal: the ask is skipped, never sent, and
+    the others are saved."""
+    base, data = live
+    b = _asks(data / "b")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a1", "no")
+    _choice(page, "a3", "yes")
+    with page.expect_navigation():
+        _press(page, "a2")
+    page.close()
+    got = _answers(b)
+    assert got["a1"]["choice"] == "no" and got["a3"]["choice"] == "yes", got
+    assert got["a2"] is None, "a blank ask was recorded"
+    assert [a for _, a in posts] == ["a1", "a3"], posts
+
+
+def test_an_ask_nobody_touched_is_not_re_sent(browser, live):
+    """Re-sending an answer re-dates it, and a reading session sees a fresh
+    answer that nobody gave. Only what changed since the page loaded is sent —
+    not the recorded answer sitting under the button that was pressed."""
+    from booth.marks import answer_pick
+
+    base, data = live
+    b = _asks(data / "b", ("a1", "a2"))
+    answer_pick(b, "a1", "yes")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a2-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a2", "no")
+    with page.expect_navigation():
+        _press(page, "a1")                      # the "Update answer" button
+    page.close()
+    assert [a for _, a in posts] == ["a2"], posts
+    assert _answers(b)["a2"]["choice"] == "no"
+
+
+def test_one_changed_ask_still_submits_as_a_plain_form(browser, live):
+    """With nothing else on the page to save, a submit is exactly what it was:
+    the browser's own form POST and its 303. The batch is an addition that
+    engages only when another ask holds unsent input."""
+    base, data = live
+    b = _asks(data / "b")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a2", "yes")
+    with page.expect_navigation():
+        _press(page, "a2")
+    page.close()
+    assert posts == [("document", "a2")], posts
+    assert _answers(b)["a2"]["choice"] == "yes"
+
+
+def test_the_asks_are_sent_in_document_order(browser, live):
+    """INV-6: the batch is an ordered collection, and its rule is the order the
+    forms sit in the document — here the REVERSE of the payload's, because the
+    author anchored a2 above a1."""
+    base, data = live
+    b = _asks(data / "b", ("a1", "a2"))
+    html = ("<!doctype html><title>r</title><body>"
+            '<div data-booth-mark="a2"></div><div data-booth-mark="a1"></div>'
+            f"{SEAM}</body>")
+    page = _open(browser, base, "b", html, b)
+    page.wait_for_selector("#bk-ask-a1-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a1", "yes")
+    _choice(page, "a2", "yes")
+    with page.expect_navigation():
+        _press(page, "a1")
+    page.close()
+    assert [a for _, a in posts] == ["a2", "a1"], posts
+
+
+def test_a_refused_ask_costs_only_itself_and_clears_nothing(browser, live):
+    """The session withdrew a2 while the operator was answering. a1 and a3 are
+    still saved — one stale ask must not cost the rest — and the page does NOT
+    reload, so the pick he made for a2 is still on screen, and the page says
+    which one did not save."""
+    from booth.marks import delete_mark
+
+    base, data = live
+    b = _asks(data / "b")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    page.evaluate("window.__same = 1")
+    _choice(page, "a1", "yes")
+    _choice(page, "a2", "yes")
+    _choice(page, "a3", "no")
+    delete_mark(b, "a2")
+    _press(page, "a3")
+    page.wait_for_selector("#bk-ask-a3-submit .bk-ask-status:not([hidden])", timeout=10000)
+    said = page.locator("#bk-ask-a3-submit .bk-ask-status").inner_text()
+    kept = page.is_checked('input[name="choice"][form="bk-ask-form-a2"][value="yes"]')
+    same = page.evaluate("window.__same === 1")
+    page.close()
+    got = _answers(b)
+    assert got["a1"]["choice"] == "yes" and got["a3"]["choice"] == "no", got
+    assert "a2" not in got
+    assert same, "a failed batch reloaded the page and cleared what was entered"
+    assert kept, "the refused ask's pick was cleared"
+    assert "a2" in said and "2 of 3" in said, said
+
+
+def test_an_authors_own_form_is_never_taken_over(browser, live):
+    """The batch listens to the forms the SCRIPT mounted. An author's own form
+    on the same report — a search box, say — submits as the author wrote it,
+    even while the operator has unsent picks, and nothing of ours is sent."""
+    base, data = live
+    b = _asks(data / "b", ("a1", "a2"))
+    html = ("<!doctype html><title>r</title><body>"
+            # an id with OUR prefix, so the mounted-root check is what excludes
+            # it, not the selector (kimi, hulda: the prefix alone hid the guard)
+            '<form id="bk-ask-form-theirs" action="/b/b/" method="get"><input name="q" value="x">'
+            '<button id="go">go</button></form>'
+            f"{SEAM}</body>")
+    page = _open(browser, base, "b", html, b)
+    page.wait_for_selector("#bk-ask-a2-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a1", "yes")
+    _choice(page, "a2", "no")
+    with page.expect_navigation():
+        page.click("#go")
+    url = page.url
+    page.close()
+    assert posts == [], posts
+    assert url.endswith("?q=x"), url
+    assert _answers(b) == {"a1": None, "a2": None}
+
+
+# ---- the flight window (heid bug-hunt, 2026-09-27, 4 of 4 arms) --------------
+#
+# The batch reads every form at the press, but the page stays live for the
+# whole flight. Every guard that protects that window survived its mutation,
+# because no test pressed or edited inside one. These do: each POST's reply is
+# held 700ms in the CLIENT, so the window is wide enough to act in on purpose.
+
+_HOLD_POSTS = """
+(function () {
+  var real = window.fetch;
+  window.fetch = function (u, o) {
+    var p = real.apply(this, arguments);
+    if (o && o.method === 'POST') {
+      return p.then(function (r) {
+        return new Promise(function (res) { setTimeout(function () { res(r); }, 700); });
+      });
+    }
+    return p;
+  };
+})();
+"""
+
+
+def _open_held(browser, base, name, html, booth):
+    (booth / "index.html").write_text(html, encoding="utf-8")
+    page = browser.new_page()
+    page.add_init_script(_HOLD_POSTS)
+    page.goto(f"{base}/b/{name}/", wait_until="networkidle")
+    return page
+
+
+def test_a_press_inside_the_flight_never_fires_a_native_post(browser, live):
+    """hulda: press a blank ask's button (a batch of the OTHER one starts), then
+    that other ask's own button. No other form is dirty any more from its point
+    of view, so it used to fall through to the browser — a native POST of an
+    answer already in flight, and a navigation racing the batch."""
+    base, data = live
+    b = _asks(data / "b", ("a1", "a2"))
+    page = _open_held(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a2-submit")
+    posts = _watch_posts(page)
+    _choice(page, "a2", "yes")
+    with page.expect_navigation(timeout=10000):
+        _press(page, "a1")
+        page.wait_for_timeout(150)
+        _press(page, "a2")
+    page.close()
+    assert posts == [("fetch", "a2")], posts
+
+
+def test_input_made_during_the_flight_is_kept_and_saved_on_the_next_press(browser, live):
+    """All four arms: a successful batch reloaded with no carry, so a pick made
+    while it was in flight vanished. Now the page stays when anything changed
+    after the press, says so, and the next press sends ONLY that — the saved
+    answers are not sent again (their baseline moved to what the server took)."""
+    base, data = live
+    b = _asks(data / "b")
+    page = _open_held(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    page.evaluate("window.__same = 1")
+    posts = _watch_posts(page)
+    _choice(page, "a1", "yes")
+    _choice(page, "a2", "no")
+    _press(page, "a1")
+    page.wait_for_timeout(150)
+    _choice(page, "a3", "no")                        # mid-flight
+    page.wait_for_selector("#bk-ask-a1-submit .bk-ask-status:not([hidden])", timeout=10000)
+    said = page.locator("#bk-ask-a1-submit .bk-ask-status").inner_text()
+    kept = page.is_checked('input[name="choice"][form="bk-ask-form-a3"][value="no"]')
+    same = page.evaluate("window.__same === 1")
+    with page.expect_navigation(timeout=10000):
+        _press(page, "a3")
+    page.close()
+    assert same and kept, (same, kept)
+    assert "not saved yet" in said, said
+    assert [a for _, a in posts] == ["a1", "a2", "a3"], posts
+    assert {k: v["choice"] for k, v in _answers(b).items()} == {"a1": "yes", "a2": "no", "a3": "no"}
+
+
+def test_a_retry_after_a_refusal_sends_only_what_did_not_save(browser, live):
+    """groa, hulda: after a partial failure the saved forms still read as dirty,
+    so the retry re-POSTed them and re-dated answers nobody changed."""
+    from booth.marks import delete_mark
+
+    base, data = live
+    b = _asks(data / "b")
+    page = _open(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a3-submit")
+    for mid in ("a1", "a2", "a3"):
+        _choice(page, mid, "yes")
+    delete_mark(b, "a2")
+    posts = _watch_posts(page)
+    _press(page, "a3")
+    page.wait_for_selector("#bk-ask-a3-submit .bk-ask-status:not([hidden])", timeout=10000)
+    page.evaluate("document.querySelector('#bk-ask-a3-submit .bk-ask-status').hidden = true")
+    _press(page, "a3")
+    page.wait_for_selector("#bk-ask-a3-submit .bk-ask-status:not([hidden])", timeout=10000)
+    page.close()
+    assert [a for _, a in posts] == ["a1", "a2", "a3", "a2"], posts
+
+
+def test_the_empty_status_line_stays_hidden_under_host_css(browser, live):
+    """groa, regin: `p{display:block}` in the author's sheet outranks the UA's
+    own [hidden]; the embed restates it at class specificity."""
+    base, data = live
+    b = _asks(data / "b", ("a1",))
+    html = ("<!doctype html><title>r</title><style>p{display:block}</style>"
+            f"<body><h1>R</h1>{SEAM}</body>")
+    page = _open(browser, base, "b", html, b)
+    page.wait_for_selector("#bk-ask-a1-submit")
+    shown = page.evaluate(
+        "getComputedStyle(document.querySelector('#bk-ask-a1-submit .bk-ask-status')).display")
+    page.close()
+    assert shown == "none", shown
+
+
+def test_a_refusal_blocks_the_reload_even_when_nothing_reads_dirty(browser, live):
+    """hulda: the reload waited only on "nothing dirty". A refused form the
+    operator then set back to its first value reads clean — so the page
+    reloaded over a failure and never said it. A refusal blocks the reload on
+    its own."""
+    from booth.marks import answer_pick, delete_mark
+
+    base, data = live
+    b = _asks(data / "b", ("a1", "a2"))
+    answer_pick(b, "a1", "yes")
+    answer_pick(b, "a2", "yes")
+    page = _open_held(browser, base, "b", PLAIN, b)
+    page.wait_for_selector("#bk-ask-a2-submit")
+    page.evaluate("window.__same = 1")
+    _choice(page, "a1", "no")
+    _choice(page, "a2", "no")
+    delete_mark(b, "a1")
+    _press(page, "a2")
+    page.wait_for_timeout(150)
+    _choice(page, "a1", "yes")                  # back to its first value, mid-flight
+    page.wait_for_timeout(2500)
+    same = page.evaluate("window.__same === 1")
+    shown = page.evaluate(
+        "!document.querySelector('#bk-ask-a2-submit .bk-ask-status').hidden")
+    page.close()
+    assert same, "the page reloaded over a refused POST"
+    assert shown, "the refusal was never said"
diff --git a/tests/test_flow_browser.py b/tests/test_flow_browser.py
index 18f5882..de616de 100644
--- a/tests/test_flow_browser.py
+++ b/tests/test_flow_browser.py
@@ -140,6 +140,7 @@ def test_a_failed_save_says_so_reloads_and_never_re_posts(browser, live):
     page = browser.new_page()
     page.on("request", lambda r: posts.append(r.url) if r.method == "POST" else None)
     page.goto(f"{base}/b/g/", wait_until="networkidle")
+    page.evaluate("window.__same = 1")
     (b / ".marks.json").write_text("{damaged")
     page.locator('figure.item[data-item="01.png"] .flagtoggle button').click()
     # the reader is TOLD before the page goes (Wren, hulda: nothing asserted it)
@@ -147,8 +148,12 @@ def test_a_failed_save_says_so_reloads_and_never_re_posts(browser, live):
     said = page.locator('[data-region="status"]').inner_text()
     page.wait_for_load_state("networkidle")
     page.wait_for_timeout(1500)          # past the reload
+    # ...and it DID reload, which the name promises and nothing asserted (heid
+    # bug-hunt 2026-09-27, hulda: deleting the reload survived this test)
+    reloaded = page.evaluate("window.__same !== 1")
     page.close()
     assert "Could not save in place" in said
+    assert reloaded, "the failure path never reloaded"
     assert len(posts) == 1, f"re-POSTed: {posts}"
 
 
@@ -1499,3 +1504,305 @@ def test_a_classic_scrollbar_is_neither_under_an_arrow_nor_a_pan(browser, live):
     assert g["gutter"] >= 15, ("the forced classic scrollbar is not in effect", g)
     assert g["next_right"] <= g["client_right"] - 8 + 1, g
     assert left == 800, left
+
+
+# ---- one submit saves every changed pick on the page (2026-09-27) -------------
+#
+# The Booth-page half of the operator's report (the verbatim half is in
+# test_embed_browser.py). Here an unsent pick SURVIVED another save — C3 carries
+# dirty controls — but it was never SAVED, and pressing the submit of a blank
+# pick was a 400, whose failure path reloads and wipes every one of them.
+# Contract: r2_flow C3, "Several picks at once".
+
+
+def _picks(b, ids=("a1", "a2", "a3")):
+    from booth.marks import declare_pick
+    for mid in ids:
+        declare_pick(b, mid, {"prompt": f"About {mid}?", "options": ["yes", "no"]})
+
+
+def _chosen(b):
+    from booth.marks import marks_for
+    return {m.id: (m.answer or {}).get("choice") for m in marks_for(b) if m.shape == "pick"}
+
+
+def test_one_submit_on_the_marks_page_saves_every_changed_pick(browser, live):
+    base, root = live
+    b = _set(root, 1)
+    _picks(b)
+    page = browser.new_page()
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.evaluate("window.__same = 1")
+    page.check('#mark-a1 input[type=radio][value="yes"]')
+    page.check('#mark-a2 input[type=radio][value="no"]')
+    page.check('#mark-a3 input[type=radio][value="yes"]')
+    page.locator("#mark-a3 .mark-submit").click()
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 3", timeout=10000)
+    same = page.evaluate("window.__same === 1")
+    page.close()
+    assert _chosen(b) == {"a1": "yes", "a2": "no", "a3": "yes"}
+    assert same, "the save reloaded instead of landing in place"
+
+
+def test_pressing_a_blank_picks_submit_saves_the_others_and_skips_it(browser, live):
+    """On the lightbox, from the verdict aside. The pressed pick is blank: it
+    is not sent (a 400, whose failure reload wiped the rest), the other two
+    are, and the page stays put."""
+    base, root = live
+    b = _set(root, 3)
+    _picks(b)
+    posts = []
+    page = browser.new_page(viewport={"width": 1400, "height": 900})
+    page.on("request", lambda r: posts.append(r.post_data) if r.method == "POST" else None)
+    page.goto(f"{base}/b/g/", wait_until="networkidle")
+    page.evaluate("window.__same = 1")
+    page.check('.verdict #mark-a1 input[type=radio][value="no"]')
+    page.check('.verdict #mark-a3 input[type=radio][value="no"]')
+    page.locator(".verdict #mark-a2 .mark-submit").click()
+    page.wait_for_function(
+        "document.querySelectorAll('.verdict .mark-pick.is-answered').length === 2", timeout=10000)
+    same = page.evaluate("window.__same === 1")
+    page.close()
+    assert _chosen(b) == {"a1": "no", "a2": None, "a3": "no"}
+    assert [p.split("&")[0] for p in posts] == ["ask=a1", "ask=a3"], posts
+    assert same
+
+
+def test_a_refused_pick_in_a_batch_costs_only_itself_and_clears_nothing(browser, live):
+    """a2's POST never reaches the server (the network drops it). a1 and a3
+    still land, nothing is sent twice, and — heid bug-hunt, 3 of 4 arms — the
+    page does NOT reload: the saved picks come back in place, a2's pick and a
+    half-typed note are still on screen, and the status line names a2."""
+    base, root = live
+    b = _set(root, 1)
+    _picks(b)
+    posts = []
+    page = browser.new_page()
+    page.on("request", lambda r: posts.append(r.post_data) if r.method == "POST" else None)
+    page.route("**/b/g/answer", lambda route: route.abort()
+               if "ask=a2" in (route.request.post_data or "") else route.continue_())
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.evaluate("window.__same = 1")
+    for mid in ("a1", "a2", "a3"):
+        page.check(f'#mark-{mid} input[type=radio][value="yes"]')
+    page.locator(".mark-add textarea").fill("half a thought")
+    page.locator("#mark-a1 .mark-submit").click()
+    page.wait_for_selector('[data-region="status"]:not([hidden])', timeout=10000)
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 2", timeout=10000)
+    said = page.locator('[data-region="status"]').inner_text()
+    page.wait_for_timeout(1500)            # past where a reload would have been
+    same = page.evaluate("window.__same === 1")
+    kept = page.is_checked('#mark-a2 input[type=radio][value="yes"]')
+    draft = page.locator(".mark-add textarea").input_value()
+    page.close()
+    assert same, "a refused batch reloaded the page"
+    assert kept and draft == "half a thought", (kept, draft)
+    assert "a2" in said and "2 of 3" in said, said
+    assert _chosen(b) == {"a1": "yes", "a2": None, "a3": "yes"}
+    assert [p.split("&")[0] for p in posts] == ["ask=a1", "ask=a2", "ask=a3"], posts
+
+
+# Every POST's reply held 700ms in the CLIENT, so a second press can land
+# inside the flight on purpose rather than by luck.
+_HOLD_POSTS = """
+(function () {
+  var real = window.fetch;
+  window.fetch = function (u, o) {
+    var p = real.apply(this, arguments);
+    if (o && o.method === 'POST') {
+      return p.then(function (r) {
+        return new Promise(function (res) { setTimeout(function () { res(r); }, 700); });
+      });
+    }
+    return p;
+  };
+})();
+"""
+
+
+def test_pressing_a_clean_pick_during_a_batch_sends_nothing(browser, live):
+    """kimi: a blank pick pressed while a batch was in flight found every
+    dirty form busy, fell to the one-form path, POSTed a blank and got the 400
+    — 'Could not save in place' at the moment the save was succeeding, and a
+    reload. The batch in flight already covers it: the press is a no-op."""
+    base, root = live
+    b = _set(root, 1)
+    _picks(b)
+    posts = []
+    page = browser.new_page()
+    page.add_init_script(_HOLD_POSTS)
+    page.on("request", lambda r: posts.append(r.post_data) if r.method == "POST" else None)
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.evaluate("window.__same = 1")
+    page.check('#mark-a1 input[type=radio][value="yes"]')
+    page.check('#mark-a3 input[type=radio][value="no"]')
+    page.locator("#mark-a1 .mark-submit").click()
+    page.wait_for_timeout(150)
+    page.locator("#mark-a2 .mark-submit").click()
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 2", timeout=10000)
+    page.wait_for_timeout(1200)
+    status_hidden = page.evaluate("document.querySelector('[data-region=\"status\"]').hidden")
+    same = page.evaluate("window.__same === 1")
+    page.close()
+    assert [p.split("&")[0] for p in posts] == ["ask=a1", "ask=a3"], posts
+    assert status_hidden and same, (status_hidden, same)
+
+
+# The first page refresh after a save held 1s, so a batch queues behind a flag
+# and the flag's swap lands while the batch is still waiting its turn.
+_HOLD_FIRST_REFRESH_AND_POSTS = _HOLD_FIRST_REFRESH + _HOLD_POSTS
+
+
+def test_a_pick_in_flight_stays_in_flight_across_another_saves_swap(browser, live):
+    """hulda: the in-flight mark lived on the DOM node, and a queued flag's
+    swap replaced the node with an unmarked copy — so a second press built a
+    second batch and POSTed the same answers twice. In flight is now keyed by
+    the form's identity, which survives a swap."""
+    base, root = live
+    b = _set(root, 2)
+    _picks(b, ("a1", "a2"))
+    posts = []
+    page = browser.new_page(viewport={"width": 1400, "height": 900})
+    page.add_init_script(_HOLD_FIRST_REFRESH_AND_POSTS)
+    page.on("request", lambda r: posts.append(r.post_data) if r.method == "POST" else None)
+    page.goto(f"{base}/b/g/", wait_until="networkidle")
+    page.locator('figure.item[data-item="01.png"] .flagtoggle button').click()
+    page.check('.verdict #mark-a1 input[type=radio][value="yes"]')
+    page.check('.verdict #mark-a2 input[type=radio][value="no"]')
+    page.locator(".verdict #mark-a1 .mark-submit").click()
+    page.wait_for_selector('figure.item.is-flagged[data-item="01.png"]', timeout=10000)
+    page.locator(".verdict #mark-a1 .mark-submit").click()     # a fresh node, same form
+    page.wait_for_function(
+        "document.querySelectorAll('.verdict .mark-pick.is-answered').length === 2", timeout=15000)
+    page.wait_for_timeout(2500)
+    page.close()
+    asks = [p.split("&")[0] for p in posts if p.startswith("ask=")]
+    assert asks == ["ask=a1", "ask=a2"], posts
+
+
+def test_a_later_save_clears_a_stale_not_saved_line(browser, live):
+    """A partial batch's "Not saved: a2" stays on screen until something saves
+    again — and then it must go, or it reads as current after a2 has saved."""
+    base, root = live
+    b = _set(root, 1)
+    _picks(b, ("a1", "a2"))
+    drop = {"a2": True}
+    page = browser.new_page()
+    page.route("**/b/g/answer", lambda route: route.abort()
+               if drop["a2"] and "ask=a2" in (route.request.post_data or "")
+               else route.continue_())
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.check('#mark-a1 input[type=radio][value="yes"]')
+    page.check('#mark-a2 input[type=radio][value="no"]')
+    page.locator("#mark-a1 .mark-submit").click()
+    page.wait_for_selector('[data-region="status"]:not([hidden])', timeout=10000)
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 1", timeout=10000)
+    drop["a2"] = False
+    page.locator("#mark-a2 .mark-submit").click()
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 2", timeout=10000)
+    hidden = page.evaluate("document.querySelector('[data-region=\"status\"]').hidden")
+    page.close()
+    assert hidden, "the stale 'not saved' line outlived the save that fixed it"
+    assert _chosen(b) == {"a1": "yes", "a2": "no"}
+
+
+# ---- round two of the flight window (heid bug-hunt, hulda, 2026-09-27) --------
+
+
+def test_a_saved_notes_box_comes_back_empty(browser, live):
+    """The form just sent comes back as the server renders it: a saved note's
+    box is EMPTY again. Nothing asserted it, so dropping the sent-form rule from
+    carry() survived its mutation (hulda) — and a box that kept its text would
+    post the same note twice on the next press."""
+    from booth.marks import marks_for
+    base, root = live
+    b = _set(root, 1)
+    page = browser.new_page()
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.locator(".mark-add textarea").fill("said once")
+    page.locator(".mark-add button").click()
+    page.wait_for_selector(".mark-note", timeout=10000)
+    left = page.locator(".mark-add textarea").input_value()
+    page.close()
+    assert left == "", f"the saved note's text came back as a draft: {left!r}"
+    assert [m.text for m in marks_for(b) if m.shape == "note"] == ["said once"]
+
+
+def test_a_change_to_a_sent_pick_during_the_flight_is_kept(browser, live):
+    """hulda #1: carry() skipped EVERY control of a sent form, so a pick changed
+    after the press came back as the server's copy of the earlier one. A sent
+    form that changed since the press now carries like any unsent form."""
+    base, root = live
+    b = _set(root, 1)
+    _picks(b, ("a1", "a2"))
+    page = browser.new_page()
+    page.add_init_script(_HOLD_POSTS)
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    page.check('#mark-a1 input[type=radio][value="yes"]')
+    page.check('#mark-a2 input[type=radio][value="yes"]')
+    page.locator("#mark-a1 .mark-submit").click()
+    page.wait_for_timeout(150)
+    page.check('#mark-a1 input[type=radio][value="no"]')      # after the press
+    page.wait_for_function(
+        "document.querySelectorAll('.mark-pick.is-answered').length === 2", timeout=10000)
+    page.wait_for_timeout(300)
+    showing = page.is_checked('#mark-a1 input[type=radio][value="no"]')
+    page.close()
+    assert _chosen(b) == {"a1": "yes", "a2": "yes"}
+    assert showing, "the change made after the press was replaced by the saved copy"
+
+
+def test_a_queued_note_is_not_carried_back_as_a_draft(browser, live):
+    """hulda #3: a note queued behind a flag. The flag's swap replaced the note
+    form, so the note's own refresh named a detached node, the sent-form rule
+    missed the live one, and the saved text came back as a draft — one more
+    press and it posted twice. Sent forms now match by identity."""
+    from booth.marks import marks_for
+    base, root = live
+    b = _set(root, 2)
+    page = browser.new_page(viewport={"width": 1400, "height": 900})
+    page.add_init_script(_HOLD_FIRST_REFRESH)
+    page.goto(f"{base}/b/g/", wait_until="networkidle")
+    page.locator('figure.item[data-item="01.png"] .flagtoggle button').click()
+    page.locator(".verdict .mark-add textarea").fill("queued once")
+    page.locator(".verdict .mark-add button").click()
+    page.wait_for_selector(".verdict .mark-note", timeout=10000)
+    page.wait_for_timeout(1500)
+    left = page.locator(".verdict .mark-add textarea").input_value()
+    page.close()
+    assert left == "", f"the saved note came back as a draft: {left!r}"
+    assert [m.text for m in marks_for(b) if m.shape == "note"] == ["queued once"]
+
+
+def test_a_batch_whose_refresh_fails_keeps_the_page(browser, live):
+    """hulda #4, and heid's H1: every pick saved, then the page GET failed —
+    and the batch took C3's reload, taking an unsent note with it. A batch
+    never reloads: it says the page could not be refreshed."""
+    base, root = live
+    b = _set(root, 1)
+    _picks(b, ("a1", "a2"))
+    gate = {"on": False}
+    page = browser.new_page()
+    page.route("**/b/g/marks", lambda route: route.abort()
+               if gate["on"] and route.request.method == "GET" else route.continue_())
+    page.goto(f"{base}/b/g/marks", wait_until="networkidle")
+    gate["on"] = True
+    page.evaluate("window.__same = 1")
+    page.check('#mark-a1 input[type=radio][value="yes"]')
+    page.check('#mark-a2 input[type=radio][value="no"]')
+    page.locator(".mark-add textarea").fill("not sent")
+    page.locator("#mark-a1 .mark-submit").click()
+    page.wait_for_selector('[data-region="status"]:not([hidden])', timeout=10000)
+    page.wait_for_timeout(1500)
+    said = page.locator('[data-region="status"]').inner_text()
+    same = page.evaluate("window.__same === 1")
+    draft = page.locator(".mark-add textarea").input_value()
+    page.close()
+    assert _chosen(b) == {"a1": "yes", "a2": "no"}
+    assert same and draft == "not sent", (same, draft)
+    assert "reload" in said.lower(), said