diff --git a/src/did/+did2/+convert/+migrators_j/Contents.m b/src/did/+did2/+convert/+migrators_j/Contents.m
index 77e7672f..a247544e 100644
--- a/src/did/+did2/+convert/+migrators_j/Contents.m
+++ b/src/did/+did2/+convert/+migrators_j/Contents.m
@@ -965,8 +965,12 @@
 %         Deferred: stimulus_bath migrates to a formulation-valued manipulation in the
 %         NDI layer / the coarse resolveDeferredBaths pass.
 %     stimulus_parameter
-%         Brainstorm-J migrator: did_v1 stimulus_parameter -- a PASSTHROUGH that
-%         normalises the template's own untypeable placeholder.
+%         Brainstorm-J migrator: did_v1 stimulus_parameter -- carried whole as v1
+%         (since 2026-10-29, [one folder, final classes only]; see the end of the
+%         function). Until then it was a PASSTHROUGH that normalised the template's
+%         own untypeable placeholder; the history below explains that repair, which a
+%         carried document no longer needs (it is not read against V2). A non-empty
+%         char or a non-numeric `value` still errors.
 %     stimulus_presentation
 %         Brainstorm-J migrator: did_v1 stimulus_presentation -- the decomposition is
 %         DONE BY THE NDI SECOND PASS; pass 1 is a GUARDED PASSTHROUGH that carries
diff --git a/src/did/+did2/+convert/+migrators_j/daqmetadatareader.m b/src/did/+did2/+convert/+migrators_j/daqmetadatareader.m
index 21daa370..6069bd36 100644
--- a/src/did/+did2/+convert/+migrators_j/daqmetadatareader.m
+++ b/src/did/+did2/+convert/+migrators_j/daqmetadatareader.m
@@ -134,6 +134,14 @@
 filePattern = jGetChar(blk, 'tab_separated_file_parameter');
 
 if isempty(implClass) && isempty(filePattern)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-129 covers it.
+    % carry-register: carry-129
+    did2.convert.reportNote('carried_as_v1', 'daqmetadatareader', jCarryId(src), ...
+        '[carry-129] malformed: no reader class and no file parameter, so nothing to declare ([passthrough stays v1])');
     bodies = {src};         % see THE GUARD above
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/daqreader.m b/src/did/+did2/+convert/+migrators_j/daqreader.m
index 72e50a28..23e9fdfc 100644
--- a/src/did/+did2/+convert/+migrators_j/daqreader.m
+++ b/src/did/+did2/+convert/+migrators_j/daqreader.m
@@ -151,6 +151,14 @@
 % a body with NEITHER a class name NOR a reader string says nothing at all,
 % and minting entities from it would be the hollow-document defect.
 if isempty(implClass) && isempty(readerString)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-130 covers it.
+    % carry-register: carry-130
+    did2.convert.reportNote('carried_as_v1', 'daqreader', jCarryId(src), ...
+        '[carry-130] malformed: no reader class and no reader string, so nothing to declare ([passthrough stays v1])');
     bodies = {src};
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/daqsystem.m b/src/did/+did2/+convert/+migrators_j/daqsystem.m
index 57689b4c..8ab8a93c 100644
--- a/src/did/+did2/+convert/+migrators_j/daqsystem.m
+++ b/src/did/+did2/+convert/+migrators_j/daqsystem.m
@@ -313,7 +313,8 @@
     % preserved. The passthrough rescues nothing.
     %
     % IT IS LEFT AS IT IS, DELIBERATELY, and pinned by
-    % testTheResidualPassthroughIsAQuarantinePathByConstruction. NDI sets
+    % testTheResidualPassthroughIsAQuarantinePathByConstruction (since 2026-10-29
+    % testTheResidualDaqsystemIsCarriedWholeAsV1). NDI sets
     % `ndi_daqsystem_class` on every document it writes and gives every real rig
     % a filenavigator and a daqreader, so this branch is unreachable for real
     % data; relaxing the tombstone to admit a hollow document would trade a live
@@ -325,6 +326,18 @@
     % that can carry a document on its own, via the software_id edge, so it has
     % to be counted as something-to-declare or a class-name-only daqsystem would
     % fall into a passthrough that is no longer warranted.
+    %
+    % CARRIED, NOT PASSED THROUGH, since 2026-10-29: the quarantine described
+    % above came from validating the passthrough against the `daqsystem`
+    % tombstone; a carried document is v1 and is not read against V2.
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-131 covers it.
+    % carry-register: carry-131
+    did2.convert.reportNote('carried_as_v1', 'daqsystem', jCarryId(src), ...
+        '[carry-131] malformed: no daqsystem class, navigator, reader or metadata reader, so nothing to declare ([passthrough stays v1])');
     bodies = {src};
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/electrode_offset_voltage.m b/src/did/+did2/+convert/+migrators_j/electrode_offset_voltage.m
index d9916541..daa0c9fd 100644
--- a/src/did/+did2/+convert/+migrators_j/electrode_offset_voltage.m
+++ b/src/did/+did2/+convert/+migrators_j/electrode_offset_voltage.m
@@ -121,6 +121,14 @@
 
 offset = finiteScalar(blk, 'offset');
 if isempty(offset)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-133 covers it.
+    % carry-register: carry-133
+    did2.convert.reportNote('carried_as_v1', 'electrode_offset_voltage', jCarryId(src), ...
+        '[carry-133] malformed: electrode_offset_voltage.offset is not one finite number, so nothing to fold ([passthrough stays v1])');
     bodies = {src};   % nothing to fold -> carry unchanged (discovery fallback)
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/filenavigator.m b/src/did/+did2/+convert/+migrators_j/filenavigator.m
index f478a5e9..ae26bcab 100644
--- a/src/did/+did2/+convert/+migrators_j/filenavigator.m
+++ b/src/did/+did2/+convert/+migrators_j/filenavigator.m
@@ -162,6 +162,14 @@
 mapList    = jFileMatchList(rawField(blk, 'epochprobemap_fileparameters'));
 
 if isempty(implClass) && isempty(mapFormat) && isempty(dataList) && isempty(mapList)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-132 covers it.
+    % carry-register: carry-132
+    did2.convert.reportNote('carried_as_v1', 'filenavigator', jCarryId(src), ...
+        '[carry-132] malformed: no navigator class, file patterns or epoch map, so nothing to declare ([passthrough stays v1])');
     bodies = {src};         % nothing to declare -- see THE GUARD above
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/hartley_calc.m b/src/did/+did2/+convert/+migrators_j/hartley_calc.m
index 05f5fadb..cb6c9274 100644
--- a/src/did/+did2/+convert/+migrators_j/hartley_calc.m
+++ b/src/did/+did2/+convert/+migrators_j/hartley_calc.m
@@ -397,6 +397,19 @@
 % document whole as v1 (carry-100 outside the template, carry-056 inside it),
 % as it does for an unread calculator field
 dropped = [dropped, unreadFields(preBody)];
+% NO SUBJECT: jCalculation hands such a document back unchanged (a calculation
+% requires its subject). [one folder, final classes only] (Jess Haley,
+% 2026-10-29, did-schema V_eta_go_forward_class_audit.md): a retire class's V2
+% file is deleted once no v1 document passes through under it, so it is carried
+% whole as v1 with a note here ([passthrough stays v1], Jess Haley, 2026-10-24)
+% instead of passing through silently. Register row carry-138 covers it.
+if isempty(jSubjectOf(preBody, {'element_id', 'subject_id'}))
+    % carry-register: carry-138
+    did2.convert.reportNote('carried_as_v1', 'hartley_calc', sourceId(preBody), ...
+        '[carry-138] awaiting_home: no element_id or subject_id, so no calculation subject ([passthrough stays v1])');
+    bodies = {preBody};
+    return;
+end
 bodies = jCalculation(preBody, 'receptive_field_calculation', 'rf', ...
     'hartley_calc', 'receptive field', 'ndi.calc.vis.hartley', ...
     'ExtraParameters', extra, 'ExtraInputs', inputs, 'DataBody', true, ...
diff --git a/src/did/+did2/+convert/+migrators_j/openminds_element.m b/src/did/+did2/+convert/+migrators_j/openminds_element.m
index 2fa67d11..8096f9e1 100644
--- a/src/did/+did2/+convert/+migrators_j/openminds_element.m
+++ b/src/did/+did2/+convert/+migrators_j/openminds_element.m
@@ -20,7 +20,9 @@
 %   the asserted `variable`; the ontology id + label are the term `value`. It is
 %   a timeless subject_assertion (no interaction, no session anchor). 1 -> 1.
 %
-%   If no openMINDS block is present the document is carried unchanged.
+%   If no openMINDS block is present, or the block states an `openminds_id` or
+%   a `fields.description` (carry-128), the document is carried whole as v1
+%   with a `carried_as_v1` note.
 
 arguments
     preBody (1,1) struct
@@ -33,7 +35,17 @@
     block = preBody.openminds_element;
 end
 if isempty(block)
-    v2Body = {preBody};                        % no openMINDS block: carry unchanged
+    % No openMINDS block: nothing to assert, so the document is carried whole
+    % as v1 ([passthrough stays v1], Jess Haley, 2026-10-24) rather than
+    % passed through silently under the retire class `openminds_element`
+    % ([one folder, final classes only], Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md: a retire class's V2 file is deleted
+    % once no v1 document passes through under it). No register row covers a
+    % missing block yet (a question for the team).
+    % carry-register: carry-146
+    did2.convert.reportNote('carried_as_v1', 'openminds_element', sourceIdOf(preBody), ...
+        '[carry-146] malformed: no openminds block, so no term to assert ([passthrough stays v1])');
+    v2Body = {preBody};
     return;
 end
 
@@ -41,6 +53,26 @@
 
 flds = struct();
 if isfield(block, 'fields') && isstruct(block.fields); flds = block.fields; end
+
+% [openMINDS element leftovers carried] (Decided by Jess Haley, 2026-10-29,
+% did-schema V_eta_go_forward_class_audit.md): "a v1 `openminds_element`
+% holding an `openminds_id` or a `fields.description` is carried whole as v1,
+% as for openMINDS subjects ([openMINDS subject leftovers])". The term
+% assertion keeps neither, so a document stating either is carried
+% (register row carry-128).
+omId = strtrim(jGetChar(block, 'openminds_id'));
+omDescription = strtrim(jGetChar(flds, 'description'));
+if ~isempty(omId) || ~isempty(omDescription)
+    held = {};
+    if ~isempty(omId); held{end+1} = sprintf('openminds_id "%s"', omId); end
+    if ~isempty(omDescription); held{end+1} = 'fields.description'; end
+    % carry-register: carry-128
+    did2.convert.reportNote('carried_as_v1', 'openminds_element', sourceIdOf(preBody), ...
+        sprintf(['[carry-128] awaiting_home: %s (the term assertion keeps neither) ' ...
+        '([openMINDS element leftovers carried])'], strjoin(held, ' and ')));
+    v2Body = {preBody};
+    return;
+end
 node = firstNonEmpty( ...
     jGetChar(flds, 'preferredOntologyIdentifier'), ...
     jGetChar(flds, 'ontologyIdentifier'), ...
@@ -90,6 +122,13 @@
 if isempty(out); out = 'openminds entity'; end
 end
 
+function id = sourceIdOf(preBody)
+%SOURCEIDOF The v1 document's base.id, or '' (did2.convert.v1read.readBase).
+b = did2.convert.v1read.readBase(preBody);
+id = b.id;
+if ~ischar(id); id = ''; end
+end
+
 function s = firstNonEmpty(varargin)
 s = '';
 for k = 1:numel(varargin)
diff --git a/src/did/+did2/+convert/+migrators_j/private/jCarryId.m b/src/did/+did2/+convert/+migrators_j/private/jCarryId.m
new file mode 100644
index 00000000..d34976c4
--- /dev/null
+++ b/src/did/+did2/+convert/+migrators_j/private/jCarryId.m
@@ -0,0 +1,15 @@
+function id = jCarryId(body)
+%JCARRYID The v1 document's base.id for a `carried_as_v1` note ('' when it
+%   states none), read with did2.convert.v1read.readBase (a legacy
+%   `ndi_document` identity too).
+%
+%   ID = jCarryId(BODY).
+%
+%   See also did2.convert.reportNote, did2.convert.v1read.readBase.
+
+b = did2.convert.v1read.readBase(body);
+id = b.id;
+if ~ischar(id)
+    id = '';
+end
+end
diff --git a/src/did/+did2/+convert/+migrators_j/private/jModelFit.m b/src/did/+did2/+convert/+migrators_j/private/jModelFit.m
index 0ef81664..51ec07b1 100644
--- a/src/did/+did2/+convert/+migrators_j/private/jModelFit.m
+++ b/src/did/+did2/+convert/+migrators_j/private/jModelFit.m
@@ -16,8 +16,10 @@
 %   independent_variables[] / dependent_variable (the names), goodness {sse,
 %   sse_per_point}, input_field / output_field (fit_data).
 %
-%   PASSES THROUGH (returns {PREBODY}, the v1 tombstone validates it) when
-%   the document has no subject (decided: "no subject -> refused"). A missing
+%   CARRIES the document whole as v1 (returns {PREBODY} with a
+%   `carried_as_v1` note; until 2026-10-29 a silent passthrough the v1
+%   tombstone validated) when the document has no subject (decided: "no
+%   subject -> refused"). A missing
 %   operating system no longer passes it through: `operating_system_id` is
 %   OPTIONAL (Jess Haley, 2026-10-21 [calculation-os-optional]), so the edge is
 %   omitted, never invented.
@@ -25,6 +27,18 @@
 bodies = {preBody};
 subjectId = jSubjectOf(preBody, {'element_id', 'subject_id'});
 if isempty(subjectId)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once
+    % no v1 document passes through under it, so this document, refused on
+    % purpose, is carried whole as v1 with a note ([passthrough stays v1], Jess
+    % Haley, 2026-10-24) instead of passing through silently. Register rows
+    % carry-136 (fitcurve) and carry-137 (vmspikefit) cover it.
+    stamp = did2.convert.v1read.readStamp(preBody);
+    tag = '[carry-136]';
+    if strcmp(stamp.class_name, 'vmspikefit'); tag = '[carry-137]'; end
+    % carry-register: carry-136 carry-137
+    did2.convert.reportNote('carried_as_v1', stamp.class_name, jCarryId(preBody), ...
+        [tag ' awaiting_home: no element_id or subject_id, so no calculation subject ([passthrough stays v1])']);
     return;
 end
 [software, swId, execEnv] = jSoftwareFromApp(preBody, 'RequireSession', true);
diff --git a/src/did/+did2/+convert/+migrators_j/probe_geometry.m b/src/did/+did2/+convert/+migrators_j/probe_geometry.m
index 353c2e90..526ef8cc 100644
--- a/src/did/+did2/+convert/+migrators_j/probe_geometry.m
+++ b/src/did/+did2/+convert/+migrators_j/probe_geometry.m
@@ -201,6 +201,14 @@
 end
 
 if isempty(obsBodies)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-134 covers it.
+    % carry-register: carry-134
+    did2.convert.reportNote('carried_as_v1', 'probe_geometry', jCarryId(src), ...
+        '[carry-134] malformed: no per-site axis values, so nothing to fold ([passthrough stays v1])');
     bodies = {src};   % nothing to fold -> carry unchanged (discovery fallback)
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/site2channelmap.m b/src/did/+did2/+convert/+migrators_j/site2channelmap.m
index 4e01f69c..b7fe228c 100644
--- a/src/did/+did2/+convert/+migrators_j/site2channelmap.m
+++ b/src/did/+did2/+convert/+migrators_j/site2channelmap.m
@@ -75,8 +75,17 @@
 end
 [~, collision, unplaced] = jRawAccount(preBody, struct('site2channelmap', {{'map'}}), ...
     {'^probe_id$', '^probe_geometry_id$'});
-jRawCarry('site2channelmap', preBody, collision, unplaced, 'carry-100');
-bodies = {preBody};   % a passthrough, or the carry of the untouched v1 source
+if ~jRawCarry('site2channelmap', preBody, collision, unplaced, 'carry-100')
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-139 covers it.
+    % carry-register: carry-139
+    did2.convert.reportNote('carried_as_v1', 'site2channelmap', jCarryId(preBody), ...
+        '[carry-139] awaiting_home: site2channelmap has no V2 home (its map needs the probe_geometry join) ([passthrough stays v1])');
+end
+bodies = {preBody};   % the carry of the untouched v1 source
 end
 
 % ===================== small helpers =======================================
diff --git a/src/did/+did2/+convert/+migrators_j/stimulus_parameter.m b/src/did/+did2/+convert/+migrators_j/stimulus_parameter.m
index b38acb60..b4a221e6 100644
--- a/src/did/+did2/+convert/+migrators_j/stimulus_parameter.m
+++ b/src/did/+did2/+convert/+migrators_j/stimulus_parameter.m
@@ -1,6 +1,10 @@
 function bodies = stimulus_parameter(preBody)
-%STIMULUS_PARAMETER Brainstorm-J migrator: did_v1 stimulus_parameter -- a
-%   PASSTHROUGH that normalises the template's own untypeable placeholder.
+%STIMULUS_PARAMETER Brainstorm-J migrator: did_v1 stimulus_parameter -- carried
+%   whole as v1 (since 2026-10-29, [one folder, final classes only]; see the
+%   end of the function). Until then it was a PASSTHROUGH that normalised the
+%   template's own untypeable placeholder; the history below explains that
+%   repair, which a carried document no longer needs (it is not read against
+%   V2). A non-empty char or a non-numeric `value` still errors.
 %
 %   Routed from did2.convert.v1_to_v2 only when TargetVersion == 'V_eta'.
 %
@@ -137,9 +141,10 @@
         % empty DOUBLE and validates: `value` is not scalar-constrained).
     elseif ischar(v) || isstring(v)
         if isempty(strtrim(char(v)))
-            % The template's own "unset" placeholder. Absence is the only
-            % spelling of unset that validates.
-            preBody.(BLOCK) = rmfield(preBody.(BLOCK), 'value');
+            % The template's own "unset" placeholder. It used to be removed
+            % here so the passthrough validated against the V_eta tombstone;
+            % the document is now carried whole as v1 (below), which is not
+            % read against V2, so it is kept as v1 wrote it.
         else
             error('did2:convert:stimulusParameterCharValue', ...
                 ['stimulus_parameter.value carries the non-empty char ''%s'' ' ...
@@ -155,5 +160,14 @@
     end
 end
 
+% [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it, so this document, held for the stimulus
+% model (#31) and left as v1 on purpose, is carried whole as v1 with a note
+% ([passthrough stays v1], Jess Haley, 2026-10-24) instead of passing through
+% silently. Register row carry-140 covers it.
+% carry-register: carry-140
+did2.convert.reportNote('carried_as_v1', 'stimulus_parameter', jCarryId(preBody), ...
+    '[carry-140] awaiting_home: stimulus_parameter has no V2 home until the stimulus model ([passthrough stays v1])');
 bodies = {preBody};
 end
diff --git a/src/did/+did2/+convert/+migrators_j/subjectmeasurement.m b/src/did/+did2/+convert/+migrators_j/subjectmeasurement.m
index e166aa8b..71d44678 100644
--- a/src/did/+did2/+convert/+migrators_j/subjectmeasurement.m
+++ b/src/did/+did2/+convert/+migrators_j/subjectmeasurement.m
@@ -150,6 +150,14 @@
 % ---- guard 1: no subject => no observation ---------------------------------
 subjectId = dependencyValue(preBody, 'subject_id');
 if isempty(subjectId)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-144 covers it.
+    % carry-register: carry-144
+    did2.convert.reportNote('carried_as_v1', 'subjectmeasurement', jCarryId(preBody), ...
+        '[carry-144] malformed: no subject_id, so no observation ([passthrough stays v1])');
     return;
 end
 
@@ -157,6 +165,14 @@
 % `measurement` is free text and is the whole of what the statement is ABOUT.
 name = getCharField(blk, 'measurement');
 if isempty(strtrim(name))
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-144 covers it.
+    % carry-register: carry-144
+    did2.convert.reportNote('carried_as_v1', 'subjectmeasurement', jCarryId(preBody), ...
+        '[carry-144] malformed: subjectmeasurement.measurement is empty, so no variable ([passthrough stays v1])');
     return;
 end
 
@@ -164,6 +180,14 @@
 value = [];
 if isfield(blk, 'value'); value = blk.value; end
 if ~(isnumeric(value) && isscalar(value) && isfinite(value))
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-144 covers it.
+    % carry-register: carry-144
+    did2.convert.reportNote('carried_as_v1', 'subjectmeasurement', jCarryId(preBody), ...
+        '[carry-144] malformed: subjectmeasurement.value is not one finite number ([passthrough stays v1])');
     return;
 end
 
@@ -190,6 +214,18 @@
 
 % ---- guard 4: no honest leaf => carry the document through ------------------
 if isempty(folded)
+    if did2.convert.reportNote('pending', 'carried_as_v1') == 0
+        % jMeasurementFold left no note of its own (it writes one, carry-059,
+        % for a second value): it found no quantity leaf for the name
+        % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+        % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+        % v1 document passes through under it, so this document, left as v1 on purpose,
+        % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+        % 2026-10-24) instead of passing through silently. Register row carry-145 covers it.
+        % carry-register: carry-145
+        did2.convert.reportNote('carried_as_v1', 'subjectmeasurement', jCarryId(preBody), ...
+            '[carry-145] awaiting_home: no quantity kind is named by subjectmeasurement.measurement ([passthrough stays v1])');
+    end
     return;
 end
 bodies = folded;
diff --git a/src/did/+did2/+convert/+migrators_j/syncgraph.m b/src/did/+did2/+convert/+migrators_j/syncgraph.m
index 6c941c23..b43642c5 100644
--- a/src/did/+did2/+convert/+migrators_j/syncgraph.m
+++ b/src/did/+did2/+convert/+migrators_j/syncgraph.m
@@ -167,6 +167,14 @@
     sessionId = char(preBody.base.session_id);
 end
 if isempty(sessionId) || isEmptySessionSentinel(sessionId)
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-135 covers it.
+    % carry-register: carry-135
+    did2.convert.reportNote('carried_as_v1', 'syncgraph', jCarryId(src), ...
+        '[carry-135] malformed: no session id, so no policy can be anchored ([passthrough stays v1])');
     bodies = {src};          % THE GUARD -- see the header
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/vmneuralresponseresiduals.m b/src/did/+did2/+convert/+migrators_j/vmneuralresponseresiduals.m
index 45b4d32b..1c46cdf1 100644
--- a/src/did/+did2/+convert/+migrators_j/vmneuralresponseresiduals.m
+++ b/src/did/+did2/+convert/+migrators_j/vmneuralresponseresiduals.m
@@ -92,8 +92,17 @@
     'vmneuralresponseresiduals', {{'element_epochid', 'parameters', 'column_labels', ...
         'goodness_of_fit', 'total_power', 'residual_power'}}), ...
     {'^element_id$'});
-jRawCarry('vmneuralresponseresiduals', preBody, collision, unplaced, 'carry-100');
-bodies = {preBody};   % a passthrough, or the carry of the untouched v1 source
+if ~jRawCarry('vmneuralresponseresiduals', preBody, collision, unplaced, 'carry-100')
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-141 covers it.
+    % carry-register: carry-141
+    did2.convert.reportNote('carried_as_v1', 'vmneuralresponseresiduals', jCarryId(preBody), ...
+        '[carry-141] awaiting_home: vmneuralresponseresiduals has no V2 home ([passthrough stays v1])');
+end
+bodies = {preBody};   % the carry of the untouched v1 source
 end
 
 % ===================== small helpers =======================================
diff --git a/src/did/+did2/+convert/+migrators_j/vmspikefilteringparameters.m b/src/did/+did2/+convert/+migrators_j/vmspikefilteringparameters.m
index c0eb18a5..eb9924ea 100644
--- a/src/did/+did2/+convert/+migrators_j/vmspikefilteringparameters.m
+++ b/src/did/+did2/+convert/+migrators_j/vmspikefilteringparameters.m
@@ -161,6 +161,14 @@
 
 % ---------------------------------------------------------------- THE GUARD
 if hasPayload(blk, 'spiketimes')
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-143 covers it.
+    % carry-register: carry-143
+    did2.convert.reportNote('carried_as_v1', 'vmspikefilteringparameters', jCarryId(preBody), ...
+        '[carry-143] awaiting_home: vmspikefilteringparameters.spiketimes holds a result, which the method parameters do not keep ([passthrough stays v1])');
     bodies = {preBody};
     return;
 end
diff --git a/src/did/+did2/+convert/+migrators_j/vmspikesummary.m b/src/did/+did2/+convert/+migrators_j/vmspikesummary.m
index ba8637d9..d490e66f 100644
--- a/src/did/+did2/+convert/+migrators_j/vmspikesummary.m
+++ b/src/did/+did2/+convert/+migrators_j/vmspikesummary.m
@@ -103,8 +103,17 @@
         'median_presk_halfwidthmaximum', 'median_postsk_halfwidthmaximum', 'median_max_dvdt', ...
         'median_kink_index', 'slope_criterion'}}, 'epochid', {{'epochid'}}), ...
     {'^element_id$', '^spike_extraction_id$'});
-jRawCarry('vmspikesummary', preBody, collision, unplaced, 'carry-100');
-bodies = {preBody};   % a passthrough, or the carry of the untouched v1 source
+if ~jRawCarry('vmspikesummary', preBody, collision, unplaced, 'carry-100')
+    % [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+    % V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+    % v1 document passes through under it, so this document, left as v1 on purpose,
+    % is carried whole as v1 with a note ([passthrough stays v1], Jess Haley,
+    % 2026-10-24) instead of passing through silently. Register row carry-142 covers it.
+    % carry-register: carry-142
+    did2.convert.reportNote('carried_as_v1', 'vmspikesummary', jCarryId(preBody), ...
+        '[carry-142] awaiting_home: vmspikesummary has no V2 home ([passthrough stays v1])');
+end
+bodies = {preBody};   % the carry of the untouched v1 source
 end
 
 % ===================== small helpers =======================================
diff --git a/src/did/+did2/+convert/+roundtrip/openminds_element.m b/src/did/+did2/+convert/+roundtrip/openminds_element.m
index 5cdb375a..6450a716 100644
--- a/src/did/+did2/+convert/+roundtrip/openminds_element.m
+++ b/src/did/+did2/+convert/+roundtrip/openminds_element.m
@@ -5,10 +5,12 @@
 %   [V1, SPEC] = did2.convert.roundtrip.openminds_element(BODIES). BODIES is
 %   the one `assertion` about the element (its `element_id` edge becomes the
 %   assertion's entity_id); private/openmindsInverse says what is rebuilt and
-%   what is not. Unlike openminds_subject, the migrator converts EVERY
-%   openminds_element, whether or not it states an `openminds_id`, and writes
-%   the id nowhere; no decision covers that id or the other openMINDS fields
-%   for an element, so the round trip reports each one v1 states.
+%   what is not. A document stating an `openminds_id` or a
+%   `fields.description` is carried whole as v1 by the migrator (carry-128,
+%   [openMINDS element leftovers carried], Jess Haley, 2026-10-29), stamped
+%   did_v1, and never reaches this inverse (did2.convert.roundtrip.check
+%   compares the carry itself). The other openMINDS fields v1 states (a
+%   definition, a synonym, ...) are reported.
 %
 %   See also did2.convert.roundtrip.check,
 %            did2.convert.migrators_j.openminds_element.
diff --git a/src/did/+did2/+convert/+roundtrip/openminds_subject.m b/src/did/+did2/+convert/+roundtrip/openminds_subject.m
index d78c3d34..572a1080 100644
--- a/src/did/+did2/+convert/+roundtrip/openminds_subject.m
+++ b/src/did/+did2/+convert/+roundtrip/openminds_subject.m
@@ -8,7 +8,10 @@
 %   what is not. `matlab_type` is compared as notation ([openMINDS subject
 %   leftovers], Jess Haley, 2026-10-27: "`matlab_type` is notation"): any
 %   openMINDS MATLAB class name ending in the rebuilt type's name states the
-%   same type. A document the migrator carries (an `openminds_id` that does
+%   same type. A Species' `definition` and `synonym` are accepted
+%   differences ([species text is the openMINDS equivalent], Jess Haley,
+%   2026-10-29): compared by spec.same, they are reported as `equivalent`,
+%   not lost. A document the migrator carries (an `openminds_id` that does
 %   not rebuild, a lab-made strain) is stamped did_v1 and never reaches this
 %   inverse (did2.convert.roundtrip.check compares the carry itself).
 
@@ -43,7 +46,28 @@
         'fn', @(a, b) idRebuilds(node, a), 'why', ['the id is rebuilt from the ' ...
         'term through the registry''s `equivalent` list ([openMINDS subject ' ...
         'leftovers], [term equivalents recorded], Jess Haley, 2026-10-27)']);
+    if endsWith(v1.openminds.openminds_type, '/Species')
+        % [species text is the openMINDS equivalent] (Decided by Jess Haley,
+        % 2026-10-29, did-schema V_eta_go_forward_class_audit.md): "a v1
+        % openMINDS subject's species `definition` and `synonym` are not stored
+        % in V2: the term registry records the openMINDS species instance as
+        % the term's equivalent ([term equivalents recorded]), and the round
+        % trip reports the v1 text as an accepted difference". So whatever text
+        % v1 states there is accepted (reported under `equivalent`, never lost).
+        why = ['a species'' definition and synonym are openMINDS''s text for the ' ...
+            'species instance the term registry records as the term''s equivalent ' ...
+            '([species text is the openMINDS equivalent], Jess Haley, 2026-10-29)'];
+        for f = {'definition', 'synonym'}
+            spec.same(end+1) = struct('path', ['openminds.fields.' f{1}], ...
+                'fn', @(a, ~) isText(a), 'why', why);
+        end
+    end
+end
 end
+
+function tf = isText(v)
+% the v1 value is text (char or one string): the openMINDS text accepted
+tf = ischar(v) || (isstring(v) && isscalar(v));
 end
 
 function tf = idRebuilds(node, v1Id)
diff --git a/tests/+did2/+unittest/testMigratorsJ.m b/tests/+did2/+unittest/testMigratorsJ.m
index 3b85f8f1..3c4f7ad2 100644
--- a/tests/+did2/+unittest/testMigratorsJ.m
+++ b/tests/+did2/+unittest/testMigratorsJ.m
@@ -4996,6 +4996,36 @@ function testOpenmindsElementBecomesTermAssertion(testCase)
 verifyEqual(testCase, a.document_class.class_name, 'assertion');   % #88
 end
 
+function testAnOpenmindsElementWithAnIdOrADescriptionIsCarried(testCase)
+% [openMINDS element leftovers carried] (Decided by Jess Haley, 2026-10-29,
+% did-schema V_eta_go_forward_class_audit.md): "a v1 `openminds_element`
+% holding an `openminds_id` or a `fields.description` is carried whole as v1,
+% as for openMINDS subjects"; the register row is carry-128. Each alone carries.
+body = struct();
+body.document_class = struct('class_name', 'openminds_element', 'class_version', '1.0.0', ...
+    'superclasses', struct('class_name', 'base', 'class_version', '1.0.0'));
+body.depends_on = struct('name', {'element_id', 'openminds'}, 'value', {'elem_9', ''});
+body.base = struct('id', 'ome_2', 'session_id', 'sess_09', ...
+    'name', '', 'datestamp', '2024-06-01T12:00:00.000Z');
+body.openminds = struct('openminds_type', 'https://openminds.om-i.org/types/CellType', ...
+    'matlab_type', 'openminds.controlledterms.CellType', ...
+    'openminds_id', 'CellType/dc2f5dd6-5cea-477c-a035-ca2dd3e6649d', ...
+    'fields', struct('name', 'type III BNST neuron', ...
+        'preferredOntologyIdentifier', 'EMPTY:0000073', 'description', ''));
+out = runJ(body);
+verifyCarriedWholeAsV1(testCase, body, out);
+d = out.summary.migration_notes.carried_as_v1.distinct_details{1};
+verifyTrue(testCase, startsWith(d, '[carry-128] '), d);
+desc = body;
+desc.base.id = 'ome_3';
+desc.openminds.openminds_id = '';
+desc.openminds.fields.description = 'A type of neuron.';
+out = runJ(desc);
+verifyCarriedWholeAsV1(testCase, desc, out);
+d = out.summary.migration_notes.carried_as_v1.distinct_details{1};
+verifyTrue(testCase, startsWith(d, '[carry-128] '), d);
+end
+
 function testOpenmindsStimulusPassesThroughForSecondPass(testCase)
 % INVERTED, not updated. This test used to assert the document became a
 % `term_assertion` whose subject_id came from a `stimulus_id` dependency -- and
@@ -5925,6 +5955,25 @@ function testFitcurveWithNoSubjectPassesThroughInsteadOfObservingNobody(testCase
 verifyEqual(testCase, out{1}.fitcurve.fit_equation, 'gaussian');
 end
 
+function testASubjectlessFitIsCarriedWholeAsV1WithANote(testCase)
+% [one folder, final classes only] (Decided by Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): the guard above no longer passes the fit
+% through silently under its retire class; through the pipeline it is carried
+% whole as v1 with a `carried_as_v1` note ([passthrough stays v1]). No register
+% row covers it yet, so the note cites none.
+v1 = struct();
+v1.document_class = struct('class_name', 'fitcurve', 'class_version', '1.0.0', ...
+    'superclasses', struct('class_name', {'base'}, 'class_version', {'1.0.0'}));
+v1.depends_on = struct('name', {'fit_example_data_id'}, 'value', {''});
+v1.base = struct('id', 'fc_4', 'session_id', 'sess_09', 'name', '', ...
+    'datestamp', '2024-06-01T12:00:00.000Z');
+v1.fitcurve = struct('fit_equation', 'gaussian', 'fit_sse', 12.5);
+out = runJ(v1);
+verifyCarriedWholeAsV1(testCase, v1, out);
+d = out.summary.migration_notes.carried_as_v1.distinct_details{1};
+verifyTrue(testCase, contains(d, 'no calculation subject'), d);
+end
+
 function testFitcurveWithAnEmptyElementIdIsAlsoGuarded(testCase)
 % The edge being PRESENT but blank is the same defect as the edge being absent
 % -- and it is the shape the invented-empty-edge pattern produces, so it is
diff --git a/tests/+did2/+unittest/testMigratorsJClockAlignment.m b/tests/+did2/+unittest/testMigratorsJClockAlignment.m
index f146bb4d..e8947d86 100644
--- a/tests/+did2/+unittest/testMigratorsJClockAlignment.m
+++ b/tests/+did2/+unittest/testMigratorsJClockAlignment.m
@@ -371,14 +371,21 @@ function testRuleLessGraphEmitsNoConfigurationEdgeAtAll(testCase)
     did2.unittest.helpers.sessionDocumentId('sessdoc_1', 'sess_0001'));
 end
 
-function testGraphWithTheEmptySessionSentinelPassesThrough(testCase)
+function testGraphWithTheEmptySessionSentinelIsCarried(testCase)
+% INVERTED 2026-10-29 from testGraphWithTheEmptySessionSentinelPassesThrough.
+% Decided by Jess Haley, 2026-10-29 [one folder, final classes only] (did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it, so the guard now carries the document
+% whole as v1 with a `carried_as_v1` note ([passthrough stays v1]).
 % ndi.session.empty_id() is '0000000000000000_0000000000000000' -- "no specific
 % session". Pointing a REQUIRED edge at a document that by construction does not
-% exist is an orphan, not a fact, so the document passes through instead.
+% exist is an orphan, not a fact, so the document is kept as v1 instead.
 out = runJ(syncgraphFixture({'rule_a'}, '0000000000000000_0000000000000000'));
 verifyEqual(testCase, numel(out.migrated), 1);
 verifyEqual(testCase, did2.unittest.helpers.classKey(out.migrated{1}), 'syncgraph');
+verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), 'did_v1');
 verifyEqual(testCase, out.summary.unconverted_count, 1);
+verifyTrue(testCase, isfield(out.summary.migration_notes, 'carried_as_v1'), 'the carry is noted');
 end
 
 function testPolicyNeverEmitsABlankRequiredSessionEdge(testCase)
diff --git a/tests/+did2/+unittest/testMigratorsJDaqConfiguration.m b/tests/+did2/+unittest/testMigratorsJDaqConfiguration.m
index 9d504620..199b7c9e 100644
--- a/tests/+did2/+unittest/testMigratorsJDaqConfiguration.m
+++ b/tests/+did2/+unittest/testMigratorsJDaqConfiguration.m
@@ -87,7 +87,9 @@
 %   passthrough in the first place. The branch is therefore a quarantine path by
 %   construction, unreachable for real data (NDI writes the class name on every
 %   document), and pinned as such by
-%   testTheResidualPassthroughIsAQuarantinePathByConstruction.
+%   testTheResidualPassthroughIsAQuarantinePathByConstruction -- INVERTED
+%   2026-10-29 to testTheResidualDaqsystemIsCarriedWholeAsV1: the branch now
+%   carries the body whole as v1 ([one folder, final classes only]).
 %
 %   THAT CORRECTION IS WHY THE LINE BELOW MATTERS, and it stayed true for
 %   exactly one commit: nothing here had been executed when it was written, and
@@ -437,6 +439,31 @@ function testMetadataReaderWithNothingToDeclarePassesThrough(testCase)
 verifyEqual(testCase, out.summary.unconverted_count, 1);
 end
 
+function testHollowReadersAndSystemsAreCarriedWithANote(testCase)
+% [one folder, final classes only] (Decided by Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it. The nothing-to-declare branches above
+% (daqreader, daqmetadatareader, daqsystem) still hand the document back
+% unchanged, but now as a carry: stamped did_v1, with a `carried_as_v1` note
+% ([passthrough stays v1]), not a silent did_v2 passthrough.
+v0 = daqsystemV1('', {});
+v0.depends_on = struct('name', {}, 'id', {});
+bodies = {daqreaderV1(''), daqmetadatareaderV1('', ''), v0};
+for k = 1:numel(bodies)
+    out = runJ(bodies{k});
+    verifyEqual(testCase, numel(out.migrated), 1);
+    label = char(out.migrated{1}.get('document_class.class_name'));
+    verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), ...
+        'did_v1', [label ': carried']);
+    verifyTrue(testCase, isfield(out.summary.migration_notes, 'carried_as_v1'), ...
+        [label ': the carry is noted']);
+    if isfield(out.summary.migration_notes, 'carried_as_v1')
+        verifyTrue(testCase, contains(out.summary.migration_notes.carried_as_v1.distinct_details{1}, ...
+            'nothing to declare'), label);
+    end
+end
+end
+
 % ===================== daqsystem -> acquisition_system =====================
 
 function testDaqsystemClassNameNowFoldsToASoftwareEntity(testCase)
@@ -635,8 +662,20 @@ function testDaqsystemClassNameFoldValidatesUnderVEta(testCase)
 verifyEqual(testCase, classNames(out), {'acquisition_system', 'software'});
 end
 
-function testTheResidualPassthroughIsAQuarantinePathByConstruction(testCase)
-% INVERTED 2026-08-12, ON ITS FIRST EXECUTION. It was
+function testTheResidualDaqsystemIsCarriedWholeAsV1(testCase)
+% INVERTED 2026-10-29 from testTheResidualPassthroughIsAQuarantinePathByConstruction,
+% which pinned the all-empty daqsystem as a quarantine (1 quarantined, nothing
+% surviving). Decided by Jess Haley, 2026-10-29 [one folder, final classes
+% only] (did-schema V_eta_go_forward_class_audit.md): a retire class's V2 file
+% is deleted once no v1 document passes through under it, so the residual
+% branch no longer passes the body through to be validated against the
+% `daqsystem` tombstone: it carries it whole as v1 with a `carried_as_v1`
+% note ([passthrough stays v1], [generic stage is lossless]). A carried
+% document is v1 and is not read against V2, so nothing is quarantined and the
+% body survives untouched. That answers the open question the history below
+% leaves to the team ("Either it should go, or it should refuse loudly").
+%
+% THE EARLIER HISTORY, kept: INVERTED 2026-08-12, ON ITS FIRST EXECUTION. It was
 % `testDaqsystemTombstoneStillExistsForTheEmptyBody` and asserted
 % quarantine_count 0 with the body surviving as `daqsystem`. CI says 1 and
 % nothing survives, and CI is right -- the assertion was written from intent
@@ -672,12 +711,18 @@ function testTheResidualPassthroughIsAQuarantinePathByConstruction(testCase)
 v1 = daqsystemV1('', {});
 v1.depends_on = struct('name', {}, 'id', {});
 out = runJValidated(v1);
-verifyEqual(testCase, out.summary.quarantine_count, 1, ...
-    ['an all-empty daqsystem is a HOLLOW document: the tombstone requires ' ...
-     'two edges and a field it does not have. Quarantine is #37/#38 working, ' ...
-     'not a defect -- see this test''s header before changing either side.']);
-verifyEmpty(testCase, classNames(out), ...
-    'nothing survives, so the passthrough preserves nothing');
+verifyEqual(testCase, out.summary.quarantine_count, 0, ...
+    'an all-empty daqsystem is carried whole as v1, never quarantined');
+verifyEqual(testCase, numel(out.migrated), 1, 'one document: the v1 source');
+verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), ...
+    'did_v1', 'carried, not passed through as a V2 document');
+verifyEqual(testCase, char(out.migrated{1}.get('document_class.class_name')), 'daqsystem');
+notes = out.summary.migration_notes;
+verifyTrue(testCase, isfield(notes, 'carried_as_v1'), 'the carry is noted');
+if isfield(notes, 'carried_as_v1')
+    verifyTrue(testCase, contains(notes.carried_as_v1.distinct_details{1}, ...
+        'nothing to declare'), notes.carried_as_v1.distinct_details{1});
+end
 end
 
 function testTheFoldsLeaveNoEmptyRequiredEdgeAndNoFragment(testCase)
diff --git a/tests/+did2/+unittest/testMigratorsJFileNavSoftware.m b/tests/+did2/+unittest/testMigratorsJFileNavSoftware.m
index 2b1df016..b1a40e62 100644
--- a/tests/+did2/+unittest/testMigratorsJFileNavSoftware.m
+++ b/tests/+did2/+unittest/testMigratorsJFileNavSoftware.m
@@ -151,19 +151,25 @@ function testFilenavigatorEmptyListTextDeclaresNoPatterns(testCase)
 verifyEqual(testCase, blk.epoch_map_format, 'ndi.epoch.epochprobemap_daqsystem');
 end
 
-function testFilenavigatorWithNothingToDeclarePassesThrough(testCase)
-% THE GUARD (the fitcurve / openminds_stimulus / probe_geometry shape). All four
-% source fields empty means an epoch_file_pattern with no patterns, no format and
-% no software edge -- a hollow document that validates clean and says nothing.
-% The source class keeps its V_eta tombstone, so passing through loses nothing
-% and shows up in `unconverted_by_class` where it can be counted.
+function testFilenavigatorWithNothingToDeclareIsCarried(testCase)
+% INVERTED 2026-10-29 from testFilenavigatorWithNothingToDeclarePassesThrough,
+% which relied on the source class keeping its V_eta tombstone. Decided by Jess
+% Haley, 2026-10-29 [one folder, final classes only] (did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it, so the guard now carries the document
+% whole as v1 with a `carried_as_v1` note ([passthrough stays v1]).
+% THE GUARD: all four source fields empty means an epoch_file_pattern with no
+% patterns, no format and no software edge -- a hollow document. It is still
+% counted in `unconverted_by_class` (handed back unchanged).
 v1 = navBody('', '', '');
 v1.filenavigator.ndi_filenavigator_class = '';
 out = runJ(v1);
 verifyEmpty(testCase, out.quarantine);
 verifyEqual(testCase, numel(out.migrated), 1);
 verifyEqual(testCase, did2.unittest.helpers.classKey(out.migrated{1}), 'filenavigator');
+verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), 'did_v1');
 verifyEqual(testCase, out.summary.unconverted_count, 1);
+verifyTrue(testCase, isfield(out.summary.migration_notes, 'carried_as_v1'), 'the carry is noted');
 end
 
 function testFilenavigatorWithoutImplementationClassOmitsTheEdge(testCase)
diff --git a/tests/+did2/+unittest/testRawV1RoundTripIngestedEpochs.m b/tests/+did2/+unittest/testRawV1RoundTripIngestedEpochs.m
index 2b605f7e..4fd74cb9 100644
--- a/tests/+did2/+unittest/testRawV1RoundTripIngestedEpochs.m
+++ b/tests/+did2/+unittest/testRawV1RoundTripIngestedEpochs.m
@@ -25,8 +25,10 @@
 %                                         with edges set
 %   No corpus of the four zips holds any of them.
 %
-%   All four pass through pass 1 unchanged, as before; each round trips as
-%   itself. No known findings.
+%   The two readers pass through pass 1 unchanged, as before (resolveEpochProbemap
+%   folds their files); vmneuralresponseresiduals and vmspikesummary, which no
+%   pass reads, are carried whole as v1 since 2026-10-29 ([one folder, final
+%   classes only], Jess Haley). Each round trips as itself. No known findings.
 tests = functiontests(localfunctions);
 end
 
@@ -40,17 +42,35 @@ function testTheSwitchListsTheFamily(testCase)
 verifyEqual(testCase, r.summary.legacy_ndi_document.bodies_read_raw_v1, size(f, 1));
 end
 
-function testFixturesRoundTripAsPassthroughs(testCase)
+function testFixturesRoundTripAsPassthroughsOrCarries(testCase)
+% INVERTED 2026-10-29 from testFixturesRoundTripAsPassthroughs, which asserted
+% all four were silent pass-1 passthroughs stamped did_v2. Decided by Jess
+% Haley, 2026-10-29 [one folder, final classes only] (did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it. vmneuralresponseresiduals and
+% vmspikesummary (no V2 home, read by no later pass) are now carried whole as
+% v1 with a note ([passthrough stays v1]); the two readers still pass through
+% for resolveEpochProbemap, which folds their files.
 f = fixtures();
 for k = 1:size(f, 1)
     [label, v1] = f{k, :};
     out = runRaw(v1);
     verifyEmpty(testCase, out.quarantine, label);
     verifyEqual(testCase, numel(out.passthrough), 1, [label ': a pass-1 passthrough']);
-    verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), ...
-        'did_v2', [label ': a passthrough is not a carry']);
+    isCarry = any(strcmp(v1.document_class.class_name, ...
+        {'vmneuralresponseresiduals', 'vmspikesummary'}));
+    if isCarry
+        verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), ...
+            'did_v1', [label ': carried whole as v1']);
+        [carried, notes] = did2.unittest.helpers.carryNote({v1});
+        verifyTrue(testCase, carried, [label ': carried as the untouched v1 source']);
+        verifyTrue(testCase, contains(notes, 'has no V2 home'), notes);
+    else
+        verifyEqual(testCase, char(out.migrated{1}.get('document_class.schema_version')), ...
+            'did_v2', [label ': a passthrough is not a carry']);
+    end
     r = did2.convert.roundtrip.check(v1, out.migrated);
-    verifyFalse(testCase, r.carried, label);
+    verifyEqual(testCase, r.carried, isCarry, label);
     did2.unittest.helpers.verifyRoundTrip(testCase, r, label, {});
 end
 end
diff --git a/tests/+did2/+unittest/testRawV1RoundTripMeasurement.m b/tests/+did2/+unittest/testRawV1RoundTripMeasurement.m
index d45a0ca5..8c235e05 100644
--- a/tests/+did2/+unittest/testRawV1RoundTripMeasurement.m
+++ b/tests/+did2/+unittest/testRawV1RoundTripMeasurement.m
@@ -27,8 +27,11 @@
 %   the renamed path it passed through as a V2 `measurement`, a `retire` class,
 %   which NDI's last step stored as v1: the same end state). subjectmeasurement
 %   converts with its datestamp as an absolute time reference.
-%   stimulus_parameter is kept as it is (its decided target is built by no
-%   migrator yet). No known findings.
+%   stimulus_parameter is carried whole as v1 (its decided target is built by
+%   no migrator yet; INVERTED 2026-10-29 from "kept as it is", a silent
+%   pass-1 passthrough: [one folder, final classes only], Jess Haley,
+%   2026-10-29, deletes a retire class's V2 file once no v1 document passes
+%   through under it). No known findings.
 tests = functiontests(localfunctions);
 end
 
@@ -76,6 +79,30 @@ function testADateIsCarriedNotPassedThrough(testCase)
 verifyTrue(testCase, contains(notes, 'no V2 statement'), notes);
 end
 
+function testAStimulusParameterIsCarriedNotPassedThrough(testCase)
+% [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): carried whole as v1 with a note
+% ([passthrough stays v1]); no register row covers it yet
+[carried, notes] = did2.unittest.helpers.carryNote({stimulusParameter()});
+verifyTrue(testCase, carried);
+verifyTrue(testCase, contains(notes, 'stimulus_parameter has no V2 home'), notes);
+end
+
+function testASubjectMeasurementItCannotStateIsCarried(testCase)
+% the guards that used to pass the document through silently now carry it
+% with a note ([one folder, final classes only], Jess Haley, 2026-10-29)
+v1 = subjectAge();
+v1.depends_on = struct('name', 'subject_id', 'value', '');
+[carried, notes] = did2.unittest.helpers.carryNote({v1});
+verifyTrue(testCase, carried, 'no subject');
+verifyTrue(testCase, contains(notes, 'no subject_id'), notes);
+v1 = subjectAge();
+v1.subjectmeasurement.measurement = '';
+[carried, notes] = did2.unittest.helpers.carryNote({v1});
+verifyTrue(testCase, carried, 'no name');
+verifyTrue(testCase, contains(notes, 'subjectmeasurement.measurement is empty'), notes);
+end
+
 function testTheConvertedDocumentsValidate(testCase)
 f = fixtures();
 for k = 1:size(f, 1)
@@ -129,7 +156,7 @@ function testAnUnplacedDocumentFailsTheRoundTrip(testCase)
     'measurement, a term', measurementTerm(), 'converted'; ...
     'measurement, a date', measurementDate(), 'carried'; ...
     'subjectmeasurement, age with a datestamp', subjectAge(), 'converted'; ...
-    'stimulus_parameter, temptable2stimulusparameters', stimulusParameter(), 'kept'};
+    'stimulus_parameter, temptable2stimulusparameters', stimulusParameter(), 'carried'};   % INVERTED 2026-10-29 from 'kept'
 end
 
 function d = writerMeasurement(id, node, name, stringValue, numericValue)
diff --git a/tests/+did2/+unittest/testRawV1RoundTripProbe.m b/tests/+did2/+unittest/testRawV1RoundTripProbe.m
index 7884d822..01f49826 100644
--- a/tests/+did2/+unittest/testRawV1RoundTripProbe.m
+++ b/tests/+did2/+unittest/testRawV1RoundTripProbe.m
@@ -32,8 +32,13 @@
 %                       ([probe contact geometry carried]). The same document
 %                       with no shank_id (the "converted shape", DID-side)
 %                       converts and loses nothing.
-%     site2channelmap   a pass-1 passthrough (the NDI second pass joins it to
-%                       its geometry); it round trips as itself.
+%     site2channelmap   carried whole as v1 (no V2 home; nothing joins it to
+%                       its geometry yet). INVERTED 2026-10-29 from "a pass-1
+%                       passthrough": [one folder, final classes only] (Jess
+%                       Haley, 2026-10-29) deletes a retire class's V2 file once
+%                       no v1 document passes through under it, so the
+%                       migrator now carries it with a note. It round trips as
+%                       itself.
 %     electrode_offset_voltage  converts; with a temperature and with the
 %                       writer's NaN. Nothing lost.
 %   No writer-shaped document loses a field: there are no known findings for
@@ -157,6 +162,15 @@ function testAFieldTheMigratorDoesNotReadCarriesTheDocument(testCase)
 verifyCarried(testCase, v1, 'site2channelmap.headstage');
 end
 
+function testAnOffsetThatIsNotANumberIsCarriedWithANote(testCase)
+% [one folder, final classes only] (Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): the "nothing to fold" branch used to pass
+% the document through silently under the retire class; it is now carried
+% whole as v1 with a `carried_as_v1` note ([passthrough stays v1])
+v1 = writerOffset(NaN, 21.5);
+verifyCarried(testCase, v1, 'electrode_offset_voltage.offset is not one finite number');
+end
+
 function testTheWriterGeometryCarriesItsShanks(testCase)
 [carried, notes] = did2.unittest.helpers.carryNote({writerGeometry()});
 verifyTrue(testCase, carried);
@@ -237,7 +251,7 @@ function testAnUnplacedDocumentFailsTheRoundTrip(testCase)
 f = { ...
     'probe_geometry, fromStruct', writerGeometry(), 'carried'; ...
     'probe_geometry, converted shape', convertedGeometry(), 'converted'; ...
-    'site2channelmap', writerSiteMap(), 'passthrough'; ...
+    'site2channelmap', writerSiteMap(), 'carried'; ...   % INVERTED 2026-10-29 from 'passthrough' ([one folder, final classes only])
     'electrode_offset_voltage, with a temperature', writerOffset(0.012, 21.5), 'converted'; ...
     'electrode_offset_voltage, temperature not recorded (null)', writerOffset(-0.0031, NaN), 'converted'; ...
     'electrode_offset_voltage, temperature not recorded (NaN)', nanTemp, 'converted'};
diff --git a/tests/+did2/+unittest/testRawV1RoundTripSpike.m b/tests/+did2/+unittest/testRawV1RoundTripSpike.m
index cc94adf6..5c4bda0a 100644
--- a/tests/+did2/+unittest/testRawV1RoundTripSpike.m
+++ b/tests/+did2/+unittest/testRawV1RoundTripSpike.m
@@ -135,6 +135,17 @@ function testCarriesCiteTheirRows(testCase)
     'neuron_extracellular.waveform_sample_times (no mean_waveform');
 end
 
+function testAHartleyWithNoSubjectIsCarriedWithANote(testCase)
+% [one folder, final classes only] (Decided by Jess Haley, 2026-10-29, did-schema
+% V_eta_go_forward_class_audit.md): a hartley_calc naming no subject used to
+% pass through silently (jCalculation hands it back unchanged); it is now
+% carried whole as v1 with a `carried_as_v1` note ([passthrough stays v1])
+v1 = hartley();
+v1.depends_on(1).value = '';
+verifyCarriedFor(testCase, {v1, presentationOf(hartley())}, ...
+    'no element_id or subject_id, so no calculation subject');
+end
+
 function testAnotherJrclustNameCarries(testCase)
 % [jrclust default name is notation] (Decided by Jess Haley, 2026-10-27): only
 % the writer's default 'jrclust.prm' is notation; any other name on the
diff --git a/tests/+did2/+unittest/testTemplateLiteralTypeTraps.m b/tests/+did2/+unittest/testTemplateLiteralTypeTraps.m
index 490dd271..0bec6e42 100644
--- a/tests/+did2/+unittest/testTemplateLiteralTypeTraps.m
+++ b/tests/+did2/+unittest/testTemplateLiteralTypeTraps.m
@@ -69,7 +69,9 @@
 %   ---------------------------------------------------------------------
 %   NOTHING HERE ASSERTS A DISPOSITION
 %   ---------------------------------------------------------------------
-%   Both classes stay PASSTHROUGHS. `stimulus_parameter` is held for the
+%   Both classes stay PASSTHROUGHS (a `stimulus_parameter` is carried whole as
+%   v1 since 2026-10-29, [one folder, final classes only], so its placeholder is
+%   kept as written). `stimulus_parameter` is held for the
 %   stimulus model (#31) and `image_stack` / `image_stack_parameters` were taken
 %   BACK OUT of _DELETE_PHASE8 precisely so a passthrough has a schema to
 %   validate against. These tests gate the normalisation, not a fold. Since
@@ -183,27 +185,32 @@
 
 % ===================== stimulus_parameter.value ============================
 
-function testStimulusParameterEmptyValuePlaceholderIsDropped(testCase)
-% THE REPAIR. The template's `value: ""` is a char in a field NDI's own schema
-% types `double`, so validateTypeShape (cache.m:1229 -> :1383) rejects it. There
-% is no numeric substitute that invents nothing -- 0 is a recorded measurement,
-% NaN is a value the field may reject -- and absence is what "unset" means and
-% the only spelling that passes (cache.m:1212-1225).
+function testStimulusParameterEmptyValuePlaceholderIsCarriedAsWritten(testCase)
+% INVERTED 2026-10-29 from testStimulusParameterEmptyValuePlaceholderIsDropped.
+% Decided by Jess Haley, 2026-10-29 [one folder, final classes only] (did-schema
+% V_eta_go_forward_class_audit.md): a retire class's V2 file is deleted once no
+% v1 document passes through under it, so a stimulus_parameter is now carried
+% whole as v1 ([passthrough stays v1]). A carried document is not read against
+% V2, so the template's `value: ""` (a char in a field NDI's schema types
+% `double`, which validateTypeShape rejected) no longer has to be dropped to
+% get through: it is kept as v1 wrote it, never coerced.
 out = runJ(stimulusParameterBody(''));
 verifyEmpty(testCase, out.quarantine);
 d = onlyClass(testCase, out, 'stimulus_parameter');
+verifyEqual(testCase, char(d.get('document_class.schema_version')), 'did_v1');
 b = d.get('stimulus_parameter');
-verifyFalse(testCase, isfield(b, 'value'), ...
-    'the empty char placeholder must be DROPPED, not coerced');
+verifyTrue(testCase, isfield(b, 'value'), 'the placeholder is kept, not dropped');
+verifyEqual(testCase, b.value, '', 'and not coerced');
 % the two string fields are untouched: `string` accepts '' (cache.m:1349-1371)
 verifyEqual(testCase, b.ontology_name, 'NDIC:12');
 verifyEqual(testCase, b.name, 'Command temperature constant');
 end
 
 function testStimulusParameterPlaceholderDocumentValidates(testCase)
-% The assertion that matters: with the placeholder dropped the document -- which
-% is otherwise exactly what ndi.document('stimulus_parameter') produces -- gets
-% through the validator instead of quarantining on typeMismatch.
+% The assertion that matters: the document -- exactly what
+% ndi.document('stimulus_parameter') produces -- gets through the validator
+% instead of quarantining on typeMismatch. Since 2026-10-29 it is carried whole
+% as v1 ([one folder, final classes only]), which is not validated against V2.
 out = runJValidated(stimulusParameterBody(''));
 verifyEmpty(testCase, out.quarantine);
 verifyEqual(testCase, numel(out.migrated), 1);
diff --git a/tests/+did2/+unittest/testWholeBatchRoundTrip.m b/tests/+did2/+unittest/testWholeBatchRoundTrip.m
index 367681d2..0a1dffc4 100644
--- a/tests/+did2/+unittest/testWholeBatchRoundTrip.m
+++ b/tests/+did2/+unittest/testWholeBatchRoundTrip.m
@@ -169,10 +169,16 @@ function testAnIngestedManifestRoundTripsThroughItsDecomposition(testCase)
 verifyFalse(testCase, isfield(r.failures_by_class, 'epochfiles_ingested'), describe(r));
 end
 
-function testAnOpenmindsSubjectRoundTrips(testCase)
+function testAnOpenmindsSubjectRoundTripsWithTheSpeciesTextAccepted(testCase)
+% INVERTED 2026-10-29 from testAnOpenmindsSubjectRoundTrips, which asserted the
+% species' definition and synonym were LOST. Decided by Jess Haley, 2026-10-29
+% [species text is the openMINDS equivalent] (did-schema
+% V_eta_go_forward_class_audit.md): they are not stored in V2 (the registry
+% records the openMINDS species instance as the term's equivalent), and "the
+% round trip reports the v1 text as an accepted difference".
 % JH's shapes (ndi.database.fun.openMINDSobj2ndi_document): a sex states no
 % definition and round trips clean; a species states openMINDS's definition and
-% synonym, which nothing in DID holds, and the round trip reports exactly those
+% synonym, and the round trip reports exactly those as accepted (equivalent)
 sex = openmindsV1('wb_om_sex', 'openminds_subject', 'subject_id', 'BiologicalSex', ...
     'hermaphrodite', 'PATO:0001340', '', '');
 species = openmindsV1('wb_om_sp', 'openminds_subject', 'subject_id', 'Species', ...
@@ -184,10 +190,12 @@ function testAnOpenmindsSubjectRoundTrips(testCase)
 did2.unittest.helpers.verifyRoundTrip(testCase, r, 'openminds_subject (sex)', {});
 out = did2.convert.v1_to_v2({species}, 'Validate', false, 'TargetVersion', 'V_eta');
 r = did2.convert.roundtrip.check(species, out.migrated);
-verifyEqual(testCase, sort({r.lost.path}), ...
-    {'openminds.fields.definition', 'openminds.fields.synonym'});
-verifyEmpty(testCase, r.invented);
-verifyEmpty(testCase, r.unplaced);
+verifyFalse(testCase, r.carried);
+did2.unittest.helpers.verifyRoundTrip(testCase, r, 'openminds_subject (species)', {});
+verifyTrue(testCase, all(ismember({'openminds.fields.definition', ...
+    'openminds.fields.synonym'}, r.equivalent)), ...
+    'the species text is reported as an accepted difference');
+verifyTrue(testCase, r.ok);
 end
 
 function testAnOpenmindsSubjectIdIsReadBackThroughTheEquivalentList(testCase)
@@ -213,20 +221,54 @@ function testAnOpenmindsSubjectIdIsReadBackThroughTheEquivalentList(testCase)
 did2.unittest.helpers.verifyRoundTrip(testCase, r, 'openminds_subject (no id)', {});
 end
 
-function testAnOpenmindsElementReportsWhatItDoesNotKeep(testCase)
-% Dab's shape: a cell type with an openMINDS id and a description; the
-% migrator keeps neither, and the round trip says so
+function testAnOpenmindsElementWithAnIdAndADescriptionIsCarried(testCase)
+% INVERTED 2026-10-29 from testAnOpenmindsElementReportsWhatItDoesNotKeep,
+% which asserted the element was CONVERTED and its openminds_id and description
+% LOST. Decided by Jess Haley, 2026-10-29 [openMINDS element leftovers carried]
+% (did-schema V_eta_go_forward_class_audit.md): "a v1 `openminds_element`
+% holding an `openminds_id` or a `fields.description` is carried whole as v1"
+% (carry-128), so the round trip compares the carry and nothing is lost.
+% Dab's shape: a cell type with an openMINDS id and a description
 ct = openmindsV1('wb_om_ct', 'openminds_element', 'element_id', 'CellType', ...
     'type III BNST neuron', 'EMPTY:0000073', '', '');
 ct.openminds.openminds_id = 'CellType/dc2f5dd6-5cea-477c-a035-ca2dd3e6649d';
 ct.openminds.fields.description = 'A type of neuron.';
+% openminds_element is read on the renamed path, so pass 1 hands back the
+% renamed body as a passthrough and the pipeline stores the untouched v1 source
+% (v1_to_v2.m, "A CARRY IS THE V1 SOURCE"; ndi.migrate.internal.
+% carryUnchangedAsV1). Compare what is stored, as the table-row test does.
+out = did2.convert.v1_to_v2({ct}, 'Validate', false, 'TargetVersion', 'V_eta');
+verifyTrue(testCase, any(strcmp({out.passthrough.id}, 'wb_om_ct')), ...
+    'pass 1 records the element as carried through');
+r = did2.convert.roundtrip.wholeBatch({ct}, out);
+verifyEqual(testCase, r.sources_compared, 1);
+verifyEqual(testCase, r.sources_ok, 1, describe(r));
+verifyTrue(testCase, isfield(r.stored_as_v1_by_class, 'openminds_element'), ...
+    'the element is stored as its v1 source');
+% each half alone carries too
+for half = {'id', 'description'}
+    one = ct;
+    if strcmp(half{1}, 'id')
+        one.openminds.fields.description = '';
+    else
+        one.openminds.openminds_id = '';
+    end
+    out = did2.convert.v1_to_v2({one}, 'Validate', false, 'TargetVersion', 'V_eta');
+    verifyTrue(testCase, any(strcmp({out.passthrough.id}, 'wb_om_ct')), ...
+        ['carried for its ' half{1} ' alone']);
+end
+end
+
+function testAnOpenmindsElementWithNeitherConverts(testCase)
+% the other side of [openMINDS element leftovers carried]: a cell type stating
+% neither an openminds_id nor a description still converts to the term
+% assertion, and round trips clean
+ct = openmindsV1('wb_om_ct2', 'openminds_element', 'element_id', 'CellType', ...
+    'type III BNST neuron', 'EMPTY:0000073', '', '');
 out = did2.convert.v1_to_v2({ct}, 'Validate', false, 'TargetVersion', 'V_eta');
 r = did2.convert.roundtrip.check(ct, out.migrated);
 verifyFalse(testCase, r.carried);
-verifyEqual(testCase, sort({r.lost.path}), ...
-    {'openminds.fields.description', 'openminds.openminds_id'});
-verifyEmpty(testCase, r.invented);
-verifyEmpty(testCase, r.unplaced);
+did2.unittest.helpers.verifyRoundTrip(testCase, r, 'openminds_element (converted)', {});
 end
 
 % ===================== fixtures =========================================