Stated goal (from CLAUDE.md): "Extract knowledge from the tablets that can be tested and potentially turned into new modern inventions — especially metallurgical ratios, agricultural techniques, mathematical methods, trade formulas."
After analyzing 1,467 tablets across 25+ investigations, here is an honest assessment of what this corpus can and cannot provide.
Why: Proto-Elamite tablets from Susa (~3100 BCE) are administrative records — accounts of receipts, distributions, and censuses — not technical manuals. The genre of explicit production recipes in cuneiform appears much later (Yale Babylonian Culinary Tablets ~1700 BCE, medical texts, craft inscriptions).
What I tried: A systematic hunt for scale-invariant multi-ingredient ratios matching known ancient manufacturing (bronze 9:1, bread 3:1, mud brick 5:1, electrum ratios). Candidates emerged, but when examined in context, they all resolved to distribution lists (names receiving similar allocations) or catalogs (each item recorded at qty=1), not recipes.
Example that taught me this: M297:M024 = 2.3:1 across 112× scale looked promising until I read P008083: the M297 "quantity" of 225 was actually three separate personal names (a-ri₂, i-[M352~O]-e-ri₂, i-ri₂) each receiving 75 units. Not one commodity × 3 components, but three people × allocations. The ratio was an artifact of counting, not chemistry.
Why: We cannot date individual tablets within the Proto-Elamite period, so seasonal patterns cannot be reconstructed. Tablets with grain (M036) don't describe planting, sowing rates, or yield management — they record how much grain came in or went out.
Why: Same as recipes. The Meluhha reference on P008239 confirms trade existed, but the tablet doesn't record what was traded in specific ingredient ratios.
N45 = 100 in decimal context (not 120 as in bisexagesimal). This is verifiable arithmetic — 4 tablets sum exactly under this interpretation. Proto-Elamite scribes at Susa were using a decimal system where N45 acts as the "hundreds" marker. This contradicts prior assumptions and changes how ~186 tablet entries should be read.
Capacity conversion factors: N30C = 3 × N24 (confirmed by direct equation on P008923), N39B = 75 × N24 (inferred from P008003). This reveals a capacity system with base ratios 1, 3, 15, 75 (factors ×3, ×5, ×5) — not the Sumerian ×6 or ×10 conventions.
Dual-system accounting: P008140 proves that scribes maintained PARALLEL sums for decimal-counted and capacity-measured commodities within a single total line. "4(N14) 1(N01) 3(N39B)" means "41 decimal units + 225 capacity units," not 266 mixed. This is a non-trivial accounting invention that later Mesopotamian systems also used.
Tabular bookkeeping with row×column structure on 5 tablets (P008295, P008283, P008788, P008294, P008215). Column totals on the reverse. Arithmetic verified (column g of P008295: 14 = 14 perfect match). This is cross-footed bookkeeping — invented at Susa ~5000 years ago, still used today.
Standardized document templates: 15 MDP 26 tablets follow the identical format M157(header) + [body with M195+M057/M147/M139] + M288 pu₂ (total) + M009 M003~B (authorization). This is a standardized form — the precursor of every business form ever.
Verification formulas: The zu-M003~B closing mark (13 tablets) functions as an authorization signature. This is a checked-by stamp in its earliest attested form.
Production norm of 1.0 by-product per animal (standard tier): across 8 data points spanning two independent tablets, animals of type M362+M158, M362+M123B, M362+M026H, M362+M244 yield exactly 1 capacity-unit of by-product each. This is a measurable, testable norm.
Three-tier yield hierarchy (1.0 : 1.22 : 10+): matches the known biological progression of sheep : goat : cattle dairy yields. If confirmed by sign iconography, this reveals that ~5000 years ago at Susa, accountants were using yield-based animal classification — the conceptual foundation of modern livestock productivity metrics.
Three-tablet herd management system: P008294 (animal census + product types) + P008295/P008283 (productive subset + allocations). The concept of separating herd inventory from production records is still the standard in modern ranching.
Standardized capacity measures of 6.1 (= 1×N24 + 2×N30C) and 150.1 (= 2×N39B + 1×N24). These recur across many tablets in identical notation — evidence of fixed mercantile units (standard container sizes). If ever calibrated to liters by finding a physical vessel, we'd have Proto-Elamite's canonical measure.
The question "what modern inventions can we extract?" has two answers:
Specific production recipes (bronze alloy, dye mordant, brick mixture) — these are not present in this administrative corpus. The user would need:
- Craft inscriptions from workshops
- Medical/ritual recipe texts (later periods)
- Seal cylinders with production scenes
- Archaeological residue analysis
The "inventions" in this corpus are administrative technologies rather than product recipes:
- Decimal arithmetic with 100-unit marker — a calculating system
- Dual-system parallel bookkeeping — two-currency accounting
- Cross-footed tabular records — spreadsheet precursor
- Standardized document templates — business forms
- Authorization signatures — checked-by stamps
- Livestock productivity tiers — animal classification by yield
- Standard measurement units — calibrated vessels
These are real inventions. They were invented at Susa circa 3100 BCE, and every modern organization still uses descendants of them. The "modern invention" framing fits — we've just recovered management technology rather than manufacturing technology.
To extract production recipes from Proto-Elamite, we would need:
- Tablets with sign images aligned to specific material contexts (Dahl has these; we only have transcriptions)
- Archaeological context for each tablet (workshop? kitchen? temple storeroom?)
- Residue analysis of any vessels associated with specific signs
- Calibration of N24 to liters via physical measurement of surviving Proto-Elamite vessels
- Cross-referencing with Proto-Cuneiform (contemporary Sumerian texts from Uruk III) for parallel administrative practices
Without these external inputs, our text-only analysis has hit its ceiling for recipe extraction.
Testable knowledge recovered:
- 2 confirmed numerical-system corrections (N45=100, N30C=3)
- 1 new accounting convention (dual-system parallel sums)
- 1 production norm (1.0 by-product per standard animal)
- 3 tiers in a livestock yield hierarchy
- 2 standard capacity measures (6.1 and 150.1)
- 1 verified tabular-bookkeeping structure with arithmetic
- 1 confirmed long-distance trade reference (Meluhha)
Not recovered (as expected for this corpus type):
- Metallurgical recipes
- Ceramic formulas
- Dye/textile mordant ratios
- Specific agricultural timings
- Explicit production processes
The corpus is administrative. What we found IS what administrative corpora can yield: arithmetic, organizational structure, management norms, and commercial standards. These are real, testable, and directly translatable to modern equivalents.