Week 10 Session: Cross-system comparison

Four creatures filed twice, once under our own system and once under one of last week's two. 夔's two filings drop opposite halves of one line, 帝江 has a clause none of the three can hold, and 驺吾 is the one our system alone fails.

10 May 2027 · Phase 3, Take it apart and rebuild

What this week does#

Week 9 set two systems beside ours and stopped there: the Physiologus-descended bestiary, which files a creature by what it signifies, and the yōkai catalogue, whose unit is the encounter. This week files one creature under two systems at once and writes down what each filing left out.

Four creatures, four ways two filings fail to agree:

1. Two drop lists that are complements — 夔, whose record splits down the middle, each system keeping the half the other discards.
2. A clause no system here can hold — 帝江’s 是識歌舞.
3. The creature only our system fails — 驺吾, which both of last week’s systems file without effort.
4. A drop list whose entry is a date — 九尾狐, where the thing that needs recording is not a feature.

There is no fifth section ranking the three systems. Ranking requires a job named first, and that is done inside the four.

Example 1: two filings that drop opposite halves#

夔 (kuí) has one record, in 大荒東經.

File it by week 3’s method, parts and integers, and four cells come out of it: body ox, colour 蒼, horn none-stated, foot one. Everything else in the line is residue — its light is the sun and moon (其光如日月), its voice is thunder (其聲如雷), and going in and out of water it always brings wind and rain (出入水則必風雨). Three clauses, and more of the record by length than the four cells took.

Now file it by signification. The Physiologus asks what an animal means. A light that is the sun and moon, a voice that is thunder, and weather that follows the creature in and out of water make a complete entry. The ox body and the one foot are incidental to it: the shape a sign wears is the least of what a sign does.

Both filings work. Neither is short of material. Set the two drop lists side by side and they are complements — ours discards what theirs keeps, and nothing about 夔 is lost between the two.

That is not the good news it sounds like. Our card records a grey one-footed ox and says nothing about weather. Their entry records a weather sign and says nothing about feet. The two systems contradict each other nowhere, because they never make a statement about the same clause. What they disagree about is which half of the line is the creature, and neither file records that there was another half.

Example 2: the clause no system here can hold#

帝江 (dìjiāng) was week 3’s best case. Six feet, four wings: the parts vocabulary survived the body least likely to fit it.

One clause was residue then and is residue still. 是識歌舞 — it knows song and dance. Try it in each of the three files:

• Ours. Not a part, not a count, not a place. Week 3 sent it to residue, and residue is a holding area rather than a field.
• Signification. The Physiologus needs a meaning, and a meaning is supplied by a tradition of reading the animal. No tradition supplies one for a faceless sack that dances. The entry cannot be written.
• The encounter catalogue. It needs a where, a when, and what happens to the person. The record gives 天山 and stops.

Three systems, three unrelated reasons, the same clause on the floor each time. A feature that no filing can take is the one result this method produces by itself, and it is what week 11 is handed.

Example 3: the creature only our system fails#

驺吾 (zōuwú) stands in 海內北經.

Week 9 found that 珍獸, treasure-beast, is a price and not a kind. File the record three ways and the pattern from Example 1 turns over.

Signification takes it immediately: what a creature is worth to whoever owns it is a meaning, and it is the most specific thing the line says. The encounter catalogue takes it immediately too — 林氏國 is the where, riding is what happens, and 日行千里, a thousand 里 in a day, is the outcome.

Ours produces an accurate card and drops both. Size by comparison, a colour count and a proportion all file correctly, so the body is not the problem. The valuation has no field, and the use has none either, because week 5 withdrew uses as characters of the creature on the grounds that a use states what a person gets. That ruling was right and this is what it costs.

Two creatures on one page, then, and the systems change sides between them. On 夔 the two external systems split. On 驺吾 they agree with each other and ours is the outlier.

Example 4: a drop list whose entry is a date#

九尾狐 (jiǔwěihú) files easily everywhere.

A fox shape and nine tails go into our inventory without argument. Signification does better still: week 9 noted that significance is exactly what later readers supplied to this creature, so the entry almost writes itself.

Week 7 is the problem. The record names no beast, and 九尾狐 is a handle attached afterwards. So what needs filing here is not a feature at all. It is a date — which parts of the entry stand in the line and which arrived later. Our card has no field for it, the Physiologus has none, and the catalogue has none. The bestiary marks this one entry unnamedInSource, which is a flag on a single record and not a column anything else can be read against.

Run this week’s method on 九尾狐 and it returns nothing. All three systems keep the same body and drop the same missing date, so there are no opposed drop lists to compare. The method finds what systems disagree about. It is silent about what they have in common, including what they have all failed to record.

What you leave with#

Two drop lists for one creature, written before either filing is called better, and the job each system is being judged at named on the same page as the verdict.

Carry the clauses no filing could take. 帝江’s 是識歌舞 is the clearest, and 九尾狐’s missing date is the one that defeats the method rather than the systems. Week 11 leaves the 山海经 (Shānhǎijīng, the Classic of Mountains and Seas) for material nobody built these systems for, and asks which of our characters that material declines to supply.