diff --git a/src/did/+did2/+convert/+migrators_j/daqreader_epochdata_ingested.m b/src/did/+did2/+convert/+migrators_j/daqreader_epochdata_ingested.m
index 1471c1ac..faf5ef6d 100644
--- a/src/did/+did2/+convert/+migrators_j/daqreader_epochdata_ingested.m
+++ b/src/did/+did2/+convert/+migrators_j/daqreader_epochdata_ingested.m
@@ -149,7 +149,14 @@
     bodies = preBody;
     return;
 end
-bodies = [{preBody}, refs];
+% the document is kept beside its clock references, so it is emitted
+% complete: the superclass list of its class, from the schema
+% (did2.build.superclasses; [early migration pieces fixed] (2), Jess Haley,
+% 2026-10-26: "the builder produces a complete document"; before,
+% v1_to_v2's ensureClassBlocks rewrote v1's list)
+kept = preBody;
+kept.document_class.superclasses = did2.build.superclasses('daqreader_epochdata_ingested');
+bodies = [{kept}, refs];
 end
 
 % ===================== small helpers =======================================
diff --git a/src/did/+did2/+convert/+migrators_j/daqreader_image_epochdata_ingested.m b/src/did/+did2/+convert/+migrators_j/daqreader_image_epochdata_ingested.m
index e8131c08..d2e9915d 100644
--- a/src/did/+did2/+convert/+migrators_j/daqreader_image_epochdata_ingested.m
+++ b/src/did/+did2/+convert/+migrators_j/daqreader_image_epochdata_ingested.m
@@ -343,7 +343,14 @@
 % document rather than being deleted -- deciding it is redundant is the fold's
 % call to make, not this one's.
 
-bodies = [{preBody}, refs];
+% the document is kept beside its clock references, so it is emitted
+% complete: the superclass list of its class, from the schema
+% (did2.build.superclasses; [early migration pieces fixed] (2), Jess Haley,
+% 2026-10-26: "the builder produces a complete document"; before,
+% v1_to_v2's ensureClassBlocks rewrote v1's list)
+kept = preBody;
+kept.document_class.superclasses = did2.build.superclasses('daqreader_image_epochdata_ingested');
+bodies = [{kept}, refs];
 end
 
 % ===================== small helpers =======================================
diff --git a/src/did/+did2/+convert/+migrators_j/private/jDemoFold.m b/src/did/+did2/+convert/+migrators_j/private/jDemoFold.m
index 24509ede..6025fa8a 100644
--- a/src/did/+did2/+convert/+migrators_j/private/jDemoFold.m
+++ b/src/did/+did2/+convert/+migrators_j/private/jDemoFold.m
@@ -144,6 +144,10 @@
 % ---- the rename, and the consumed blocks ----
 v2Body = preBody;
 v2Body.document_class.class_name = 'demo';
+% the superclass list of `demo`, from the schema (did2.build.superclasses), so
+% the document is complete as emitted ([early migration pieces fixed] (2),
+% Jess Haley, 2026-10-26; before, v1_to_v2's ensureClassBlocks rewrote it)
+v2Body.document_class.superclasses = did2.build.superclasses('demo');
 v2Body.demo = demoBlock;
 % `demo_ndi_mock` is the concrete block of the mock class and carries no fields;
 % it is listed so the removal does not depend on ensureClassBlocks' empty-block
diff --git a/src/did/+did2/+convert/handOnSuperclasses.m b/src/did/+did2/+convert/handOnSuperclasses.m
new file mode 100644
index 00000000..1fe4edf3
--- /dev/null
+++ b/src/did/+did2/+convert/handOnSuperclasses.m
@@ -0,0 +1,51 @@
+function body = handOnSuperclasses(body, schemaCacheOverride)
+%HANDONSUPERCLASSES Write V2's superclass list on a document pass 1 hands on.
+%
+%   BODY = did2.convert.handOnSuperclasses(BODY, SCHEMACACHE) (SCHEMACACHE []
+%   for the shared cache). The one step on pass 1's hand-on path: a v1
+%   document whose migrator hands it on unchanged to a later batch pass
+%   (did2.convert.v1_to_v2, HAND_ON_CLASSES: syncrule, epochfiles_ingested,
+%   stimulus_response_scalar, stimulus_response_scalar_parameters_basic,
+%   openminds) keeps every field as v1 wrote it; only its
+%   document_class.superclasses is rewritten from v1's form ({definition}
+%   paths) into V2's ({class_name, class_version}, did2.build.superclasses),
+%   and an EMPTY property block v1 wrote for a parent class that V2 hosts no
+%   block for (an abstract parent, e.g. `stimulus_response_scalar_parameters`)
+%   is dropped. A NON-empty such block is left for validation to report.
+%
+%   Decided by Jess Haley, 2026-11-02 [handed-on classes carried at the end]
+%   (2) (did-schema docs/plans/did_v2_go_forward_class_audit.md;
+%   docs/V2_PLAN.md G98 (a)): "a small named step on the hand-on path writes
+%   V2's superclass list for the five classes pass 1 hands on unchanged ...
+%   and `ensureClassBlocks` is deleted". This is that step; it replaces what
+%   the padding did for these five (DID-matlab #364 measured: the superclass
+%   list, and one empty v1 parent block).
+%
+%   Errors when the schema does not know the class (did2.schema.cache), so a
+%   hand-on class that lost its schema is a quarantine, not a silent pass.
+%
+%   See also did2.build.superclasses, did2.convert.v1_to_v2.
+
+arguments
+    body (1,1) struct
+    schemaCacheOverride = []
+end
+
+cache = schemaCacheOverride;
+if isempty(cache)
+    cache = did2.schema.cache.shared();
+end
+className = char(body.document_class.class_name);
+body.document_class.superclasses = did2.build.superclasses(className, ...
+    'SchemaCache', cache);
+
+hosted = cache.resolvePlacement(className).blocksContributed;
+chain = [reshape(cache.superclasses(className), 1, []), {className}];
+for cls = setdiff(chain, hosted)
+    name = cls{1};
+    if isfield(body, name) && isstruct(body.(name)) ...
+            && (isempty(body.(name)) || isempty(fieldnames(body.(name))))
+        body = rmfield(body, name);
+    end
+end
+end
diff --git a/src/did/+did2/+convert/resolveValidIntervals.m b/src/did/+did2/+convert/resolveValidIntervals.m
index 2d53b34b..0b3fa99c 100644
--- a/src/did/+did2/+convert/resolveValidIntervals.m
+++ b/src/did/+did2/+convert/resolveValidIntervals.m
@@ -996,9 +996,14 @@
 if ~isempty(anchor.tolerance_seconds)
     % Written ONLY when the source clock stated a tolerance. An empty tolerance
     % cell would assert "the timeline is good to 0 s", which the source never
-    % said; with no block, ensureClassBlocks pads it.
+    % said; with none, the block is written empty here, so the document is
+    % complete as built ([early migration pieces fixed] (2), Jess Haley,
+    % 2026-10-26: "the builder produces a complete document"; before,
+    % ensureClassBlocks added it).
     ref.time_reference = struct('clock_tolerance', ...
         durationCell(anchor.tolerance_seconds));
+else
+    ref.time_reference = struct();
 end
 value = struct('clock', struct('node', '', 'name', anchor.clock), ...
     'start', durationCell(startSeconds));
diff --git a/src/did/+did2/+convert/v1_to_v2.m b/src/did/+did2/+convert/v1_to_v2.m
index 8160521c..4edb37bf 100644
--- a/src/did/+did2/+convert/v1_to_v2.m
+++ b/src/did/+did2/+convert/v1_to_v2.m
@@ -176,6 +176,21 @@
 % 2026-11-01 [last rename-pass classes]).
 PARENT_ONLY_CLASSES = {'app', 'calculator', 'epochid', 'stimulus_response', ...
     'stimulus_response_scalar_parameters'};
+% v1 classes whose migrator HANDS ON a document unchanged to a later batch
+% pass (resolveClockAlignment: syncrule; epochMint / resolveEpochProbemap:
+% epochfiles_ingested; resolveResponseParameters: stimulus_response_scalar and
+% its parameters; resolveOpenmindsCitations: openminds). Such a document gets
+% V2's superclass list from its one named step,
+% did2.convert.handOnSuperclasses (Decided by Jess Haley, 2026-11-02
+% [handed-on classes carried at the end] (2), which names those five, the
+% ones the corpora hold). daqreader_epochdata_ingested and
+% daqreader_image_epochdata_ingested are handed on the same way (to epochMint)
+% when their source names no epoch document; no corpus holds one, and the
+% padding probe on the unit fixtures found them (DID-matlab
+% claude/W-padding-probe), so they take the same step.
+HAND_ON_CLASSES = {'syncrule', 'epochfiles_ingested', 'stimulus_response_scalar', ...
+    'stimulus_response_scalar_parameters_basic', 'openminds', ...
+    'daqreader_epochdata_ingested', 'daqreader_image_epochdata_ingested'};
 
 % a migrator transcribes the v1 source's own terms, which the term
 % registry (strictMode 'TermRegistry', tenet T19) need not hold yet: the check
@@ -301,6 +316,8 @@
 % yet produce a complete document. Denominator: the converted bodies padded.
 paddedInspected = 0;
 paddedNames = {};
+handOnNames = {};
+handOnValues = [];
 paddedValues = [];
 % DIMENSION TERMS WITH NO NODE (Jess Haley 2026-10-22 [covariate details],
 % answer "B" to 5.4): a dimension's term is bound `preferred` for now, and
@@ -709,9 +726,18 @@
                     outBody.document_class.schema_version = 'did_v1';
                 end
             else
-                outBody = did2.convert.ensureClassBlocks(v2Bodies{bi}, options.SchemaCache);
+                outBody = v2Bodies{bi};
+                if passedThrough && any(strcmp(className, HAND_ON_CLASSES))
+                    % the hand-on step ([handed-on classes carried at the
+                    % end] (2)): V2's superclass list, nothing else
+                    outBody = did2.convert.handOnSuperclasses(outBody, options.SchemaCache);
+                    [handOnNames, handOnValues] = bumpClassCounter( ...
+                        handOnNames, handOnValues, className);
+                end
+                beforePadding = outBody;
+                outBody = did2.convert.ensureClassBlocks(outBody, options.SchemaCache);
                 paddedInspected = paddedInspected + 1;
-                if ~isequaln(outBody, v2Bodies{bi})
+                if ~isequaln(outBody, beforePadding)
                     [paddedNames, paddedValues] = bumpClassCounter( ...
                         paddedNames, paddedValues, className);
                 end
@@ -925,6 +951,8 @@
     'carried_failing_veta_validation_by_class', buildByClassTable(carriedUnvalNames, carriedUnvalValues), ...
     'legacy_ndi_document', legacy, ...
     'carried_v1_template_check', carriedTemplateCheck, ...
+    'hand_on_superclasses', struct('bodies_written', sum(handOnValues), ...
+        'by_class', buildByClassTable(handOnNames, handOnValues)), ...
     'class_blocks_padding', struct('bodies_inspected', paddedInspected, ...
         'bodies_changed', sum(paddedValues), ...
         'changed_by_class', buildByClassTable(paddedNames, paddedValues)), ...