diff --git a/scripts/build_vocab_labels.py b/scripts/build_vocab_labels.py
new file mode 100644
index 0000000..4bd7c9f
--- /dev/null
+++ b/scripts/build_vocab_labels.py
@@ -0,0 +1,321 @@
+#!/usr/bin/env python3
+"""
+Build vocab_labels.parquet from the SKOS TTL vocabularies that iSamples uses.
+
+The Explorer and Python notebooks need a stable lookup from vocabulary URIs
+(e.g. https://w3id.org/isample/vocabulary/sampledfeature/1.0/pasthumanoccupationsite)
+to human-readable labels (e.g. "Past human occupation site"). This script
+parses every SKOS TTL listed in scripts/generate_vocab_docs.sh, emits one row
+per (concept URI, language) pair, and writes a single parquet file.
+
+Output columns:
+    uri          str    Concept URI (vocab-form OR data-form — see uri_form)
+    uri_form     str    "vocab"   = URI as declared in the TTL
+                        "data_v1" = synthesized URI with "/1.0/" version
+                                    segment after the scheme root (the
+                                    convention used in iSamples export
+                                    records and downstream parquet files).
+    pref_label   str    skos:prefLabel (or rdfs:label fallback)
+    lang         str    BCP47 language tag, default "en"
+    scheme       str    skos:inScheme URI (or derived)
+    definition   str?   skos:definition (best-available language)
+    alt_labels   list   skos:altLabel values plus prefLabels from any
+                        cross-vocab redeclarations of the same URI.
+    source_ttl   str    URL of the TTL the canonical row came from.
+
+The dual-form (vocab + data_v1) emission is a workaround for a known
+mismatch: the vocabulary TTLs declare concepts without a version segment,
+but iSamples export records carry URIs with a "/1.0/" segment. See
+issue #148 for the full background.
+
+Issue: https://github.com/isamplesorg/isamplesorg.github.io/issues/148
+
+Usage:
+    pip install -r scripts/requirements.txt
+    python scripts/build_vocab_labels.py              # writes ./vocab_labels.parquet
+    python scripts/build_vocab_labels.py -o /tmp/v.parquet
+"""
+from __future__ import annotations
+
+import argparse
+import sys
+from pathlib import Path
+
+import pandas as pd
+import rdflib
+from rdflib.namespace import RDF, RDFS, SKOS
+
+# Keep this list in sync with scripts/generate_vocab_docs.sh.
+# When a new vocabulary is added there, add it here too.
+VOCAB_TTLS: list[str] = [
+    # Core iSamples vocabularies
+    "https://raw.githubusercontent.com/isamplesorg/vocabularies/main/vocabulary/material_type.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/vocabularies/main/vocabulary/sampled_feature_type.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/vocabularies/main/vocabulary/material_sample_object_type.ttl",
+    # Earth Science extension
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_earth_science/main/vocabulary/earthenv_material_extension_mineral_group.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_earth_science/main/vocabulary/earthenv_material_extension_rock_sediment.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_earth_science/main/vocabulary/earthenv_sampled_feature_role.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_earth_science/main/vocabulary/earthenv_materialsampleobject_type.ttl",
+    # Archaeology / OpenContext extension
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_archaeology/main/vocabulary/opencontext_material_extension.ttl",
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_archaeology/main/vocabulary/opencontext_materialsampleobjecttype.ttl",
+    # Biology extension
+    "https://raw.githubusercontent.com/isamplesorg/metadata_profile_biology/main/vocabulary/biology_sampledfeature_extension.ttl",
+]
+
+PREFERRED_LANG = "en"
+
+# When a concept URI is declared in more than one TTL, prefer the row whose
+# source TTL's URL contains one of these path fragments. The fragments are
+# matched against the concept URI: a URI containing "vocabulary/material/"
+# prefers a row from a TTL URL containing "vocabulary/material" (i.e. the
+# core material_type.ttl) over OpenContext or Earth Sci redeclarations.
+CANONICAL_TTL_HINTS: tuple[tuple[str, str], ...] = (
+    ("vocabulary/material/",                "vocabularies/main/vocabulary/material_type"),
+    ("vocabulary/sampledfeature/",          "vocabularies/main/vocabulary/sampled_feature_type"),
+    ("vocabulary/materialsampleobjecttype/", "vocabularies/main/vocabulary/material_sample_object_type"),
+    ("vocabulary/specimentype/",             "vocabularies/main/vocabulary/material_sample_object_type"),
+)
+
+
+def _data_form_uris(vocab_uri: str) -> list[str]:
+    """Synthesize the URI form(s) used in iSamples export records.
+
+    Each iSamples scheme uses its own version segment and slug-casing
+    convention (yes, really — see issue #148). Returns possibly-multiple
+    aliases when the data layer uses inconsistent casing.
+    """
+    # Biology data is inconsistent: most slugs are Title-cased (Animalia,
+    # Fungi, Plantae) but some are lowercase (bacteria, protozoa). Emit
+    # both forms so JOINs hit either variant.
+    def _bio_variants(s: str) -> list[str]:
+        if not s:
+            return []
+        title = s[:1].upper() + s[1:]
+        lower = s.lower()
+        return list(dict.fromkeys([title, lower]))
+
+    # (scheme_root, version_segment, slug_variants_fn_or_None)
+    KNOWN_ROOTS: tuple[tuple[str, str, callable | None], ...] = (
+        ("https://w3id.org/isample/vocabulary/material/",                 "1.0", None),
+        ("https://w3id.org/isample/vocabulary/sampledfeature/",           "1.0", None),
+        ("https://w3id.org/isample/vocabulary/materialsampleobjecttype/", "1.0", None),
+        ("https://w3id.org/isample/vocabulary/specimentype/",             "1.0", None),
+        # OpenContext extension uses /0.1/ rather than /1.0/.
+        ("https://w3id.org/isample/opencontext/material/",                "0.1", None),
+        ("https://w3id.org/isample/opencontext/materialsampleobjecttype/","0.1", None),
+        # Biology extension: /1.0/ + inconsistent slug casing in the data.
+        ("https://w3id.org/isample/biology/biosampledfeature/",           "1.0", _bio_variants),
+    )
+    for root, version, variants in KNOWN_ROOTS:
+        if vocab_uri.startswith(root):
+            slug = vocab_uri[len(root):]
+            # Don't re-version a URI that already has a version segment.
+            if slug.split("/", 1)[0].replace(".", "").isdigit():
+                return []
+            slugs = variants(slug) if variants is not None else [slug]
+            return [f"{root}{version}/{s}" for s in slugs]
+    return []
+
+
+def _prefers(ttl_url: str, concept_uri: str) -> int:
+    """Return a sort key — lower is more canonical for tie-breaking.
+    A TTL whose URL matches the concept URI's expected canonical TTL gets 0;
+    everything else gets 1.
+    """
+    for uri_fragment, ttl_fragment in CANONICAL_TTL_HINTS:
+        if uri_fragment in concept_uri and ttl_fragment in ttl_url:
+            return 0
+    return 1
+
+
+def _pick_definition(g: rdflib.Graph, c: rdflib.term.Node) -> str | None:
+    """Return one definition string, preferring English when present."""
+    defs = list(g.objects(c, SKOS.definition))
+    if not defs:
+        return None
+    for d in defs:
+        if getattr(d, "language", None) == PREFERRED_LANG:
+            return str(d)
+    return str(defs[0])
+
+
+def _pick_scheme(g: rdflib.Graph, c: rdflib.term.Node) -> str | None:
+    """Return the skos:inScheme URI for a concept, if declared."""
+    for s in g.objects(c, SKOS.inScheme):
+        return str(s)
+    return None
+
+
+def extract_rows(ttl_url: str) -> list[dict]:
+    g = rdflib.Graph()
+    g.parse(ttl_url, format="turtle")
+
+    rows: list[dict] = []
+    for c in g.subjects(RDF.type, SKOS.Concept):
+        uri = str(c)
+        scheme = _pick_scheme(g, c)
+        definition = _pick_definition(g, c)
+        alt_labels = sorted({str(a) for a in g.objects(c, SKOS.altLabel)})
+
+        # One row per language of skos:prefLabel; fall back to rdfs:label.
+        pref_labels = list(g.objects(c, SKOS.prefLabel))
+        if not pref_labels:
+            pref_labels = list(g.objects(c, RDFS.label))
+
+        if not pref_labels:
+            # Concept with no label at all — emit a row with NULL label so
+            # downstream JOINs at least know the URI exists.
+            rows.append({
+                "uri": uri,
+                "pref_label": None,
+                "lang": None,
+                "scheme": scheme,
+                "definition": definition,
+                "alt_labels": alt_labels,
+                "source_ttl": ttl_url,
+            })
+            continue
+
+        for lit in pref_labels:
+            rows.append({
+                "uri": uri,
+                "uri_form": "vocab",
+                "pref_label": str(lit),
+                "lang": getattr(lit, "language", None) or PREFERRED_LANG,
+                "scheme": scheme,
+                "definition": definition,
+                "alt_labels": alt_labels,
+                "source_ttl": ttl_url,
+            })
+    return rows
+
+
+def _dedupe(rows: list[dict]) -> list[dict]:
+    """Collapse cross-vocab duplicate (uri, lang) rows.
+
+    Strategy:
+      - For each (uri, lang), pick the row whose source TTL is the canonical
+        owner of that URI's scheme (see CANONICAL_TTL_HINTS).
+      - Move any losing rows' pref_labels into the survivor's alt_labels list
+        so we don't lose information.
+    """
+    from collections import defaultdict
+    groups: dict[tuple[str, str | None], list[dict]] = defaultdict(list)
+    for r in rows:
+        groups[(r["uri"], r["lang"])].append(r)
+
+    out: list[dict] = []
+    for (uri, lang), candidates in groups.items():
+        if len(candidates) == 1:
+            out.append(candidates[0])
+            continue
+        candidates.sort(key=lambda r: (_prefers(r["source_ttl"], r["uri"]), r["source_ttl"]))
+        keep = dict(candidates[0])
+        extra = []
+        for loser in candidates[1:]:
+            if loser["pref_label"] and loser["pref_label"] != keep["pref_label"]:
+                extra.append(loser["pref_label"])
+        if extra:
+            keep["alt_labels"] = sorted(set((keep.get("alt_labels") or []) + extra))
+        out.append(keep)
+    return out
+
+
+def _emit_data_form_aliases(rows: list[dict]) -> list[dict]:
+    """For each vocab-form row, emit an alias row at the /1.0/ data-form URI
+    so JOINs against iSamples export-derived URIs work without normalization.
+    """
+    aliases: list[dict] = []
+    for r in rows:
+        for data_uri in _data_form_uris(r["uri"]):
+            clone = dict(r)
+            clone["uri"] = data_uri
+            clone["uri_form"] = "data_v1"
+            aliases.append(clone)
+    return aliases
+
+
+def main(argv: list[str] | None = None) -> int:
+    ap = argparse.ArgumentParser(description=__doc__.splitlines()[1])
+    ap.add_argument(
+        "-o", "--output",
+        default="vocab_labels.parquet",
+        type=Path,
+        help="Output parquet path (default: ./vocab_labels.parquet)",
+    )
+    ap.add_argument(
+        "--also-csv",
+        action="store_true",
+        help="Also emit a sibling .csv for diff-friendly review.",
+    )
+    ap.add_argument(
+        "--allow-partial",
+        action="store_true",
+        help=(
+            "Continue and emit an artifact even if one or more TTL sources "
+            "fail to fetch/parse. Default is to fail-loud, since this "
+            "artifact is intended for publishing."
+        ),
+    )
+    args = ap.parse_args(argv)
+
+    all_rows: list[dict] = []
+    failures: list[tuple[str, str]] = []
+    for url in VOCAB_TTLS:
+        try:
+            n_before = len(all_rows)
+            all_rows.extend(extract_rows(url))
+            print(f"  {len(all_rows) - n_before:>4} rows  {url}")
+        except Exception as e:
+            print(f"WARN: failed to parse {url}: {e}", file=sys.stderr)
+            failures.append((url, str(e)))
+
+    if failures and not args.allow_partial:
+        print(
+            f"\nERROR: {len(failures)} TTL source(s) failed; refusing to "
+            f"emit a partial artifact. Pass --allow-partial to override.",
+            file=sys.stderr,
+        )
+        for url, err in failures:
+            print(f"  - {url}: {err}", file=sys.stderr)
+        return 3
+
+    if not all_rows:
+        print("ERROR: no rows extracted; aborting.", file=sys.stderr)
+        return 2
+
+    raw_count = len(all_rows)
+    all_rows = _dedupe(all_rows)
+    deduped_collapsed = raw_count - len(all_rows)
+    print(f"\nDedupe: collapsed {deduped_collapsed} cross-vocab duplicate rows.")
+
+    aliases = _emit_data_form_aliases(all_rows)
+    print(f"Aliases: emitted {len(aliases)} data-form (/1.0/) rows.")
+    all_rows.extend(aliases)
+
+    df = pd.DataFrame(all_rows)
+    # Final sanity check
+    dupes = df.duplicated(subset=["uri", "lang"], keep=False).sum()
+    if dupes:
+        print(f"WARN: {dupes} duplicate (uri, lang) rows survived dedupe", file=sys.stderr)
+
+    args.output.parent.mkdir(parents=True, exist_ok=True)
+    df.to_parquet(args.output, index=False)
+    print(f"\nWrote {len(df):,} rows → {args.output}")
+    print(f"  by uri_form: {df['uri_form'].value_counts().to_dict()}")
+    print(f"  unique URIs: {df['uri'].nunique():,}")
+    print(f"  languages:   {sorted(df['lang'].dropna().unique().tolist())}")
+    print(f"  schemes:     {df['scheme'].nunique()} distinct skos:inScheme values")
+
+    if args.also_csv:
+        csv_path = args.output.with_suffix(".csv")
+        df.to_csv(csv_path, index=False)
+        print(f"Also wrote {csv_path}")
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())
diff --git a/scripts/requirements.txt b/scripts/requirements.txt
new file mode 100644
index 0000000..67746e4
--- /dev/null
+++ b/scripts/requirements.txt
@@ -0,0 +1,8 @@
+# Dependencies for one-shot scripts in this directory (NOT the Quarto site
+# build — that uses ../requirements.txt). Install with:
+#   pip install -r scripts/requirements.txt
+
+# build_vocab_labels.py
+rdflib>=6.3
+pandas>=2.0
+pyarrow>=14