Week 1 Session: What a classification system is

What the Classic of Mountains and Seas is, what Linnaeus did that we are going to copy, Borges' fake Chinese encyclopedia as the counter-example, and what the book already sorts by.

22 February 2027 · Phase 1, Build a system that works

What this week does#

You already classify things. A folder structure, a way of sorting laundry, a mental shelf for people you have met once. Most of them you built without deciding anything, and most fail the same way: two things that belong together end up apart and you cannot say why. This course does it on purpose, and by week 12 you build a system for material nobody has handed you one for.

This week is the ground floor: what the material is, what the method is, and how to tell a system that is merely arbitrary from one that is broken. Nothing is classified today. Four sections:

1. What the book is, with one record printed whole — 狌狌, which shows every slot a record has before anything is asked of it.
2. Linnaeus’s three moves, and the assumption underneath all three that this text breaks.
3. Borges’ fake Chinese encyclopedia, the counter-example, which ends on the test you use for the rest of the semester.
4. The system the text already has — 猼訑 and 九尾狐, nine tails each, four mountains apart, never filed together.

Sections 2 and 3 carry no creature. They are the method and the counter-example, and section 4 cannot be judged without them.

What the Classic of Mountains and Seas is#

The 山海经 (Shānhǎijīng, the Classic of Mountains and Seas) is a Chinese geographical survey, compiled in pieces between roughly the fourth century BCE and the first century CE. Nobody knows who wrote it. It survives in eighteen scrolls, and it reads like an itinerary: go three hundred 里 east, there is a mountain called X, on it grows this, in its river lives that.

Among the plants and the ore and the rivers, it describes hundreds of animals. Some are ordinary. Many are not. It is the earliest written source for a good deal of Chinese mythology, but it is not written as mythology. It is written as a survey, in the flat voice of somebody recording what is where.

That flatness is why it is our material. Here is one entry, printed the way this course prints every one:

That is the whole record. A comparison to an animal you are assumed to know, a distinguishing part, a name, a place, and sometimes a consequence. No author to ask, no second observer, nothing to measure.

How Linnaeus classified#

Now the method. Before the eighteenth century a plant’s name was a diagnostic phrase, a string of Latin adjectives long enough to tell it from its neighbours. A description pretending to be a label. It grew whenever somebody found a similar plant, it differed between authors, and you could not alphabetise it.

Linnaeus made three moves. Any classification has to make all three.

He chose characters. A small set of features you can check on every specimen. How many stamens, mammary glands or not, how many toes. The choice is the whole game, and it is a choice: he sorted plants by their sexual parts, which was neither obvious nor uncontroversial at the time.

He nested them into ranks. Kingdom, phylum, class, order, family, genus, species. Each one inside the last.

He gave two-part names. Genus plus species. Homo sapiens. One name, one kind, everywhere.

What does the nesting buy? Economy.

Say Homo sapiens and you have also said primate, mammal, chordate, animal, without writing any of it down. Six statements ride along with the name. A flat list carries none of that, which is why a folder with two hundred subfolders is unusable and one with six is not.

Underneath all three moves is an assumption about evidence. Linnaeus had specimens. Where he had none he had a drawing, and where he had neither he had the written testimony of somebody who claimed to have stood next to the animal. The system takes for granted that behind every name there is a body somebody could in principle produce.

The Classic of Mountains and Seas produces no bodies. Every difficulty in the remaining eleven weeks comes from that.

Borges’ fake Chinese encyclopedia#

One counter-example, so you know what a bad system looks like.

In a 1942 essay on John Wilkins, Borges describes a Chinese encyclopedia that does not exist, the Celestial Emporium of Benevolent Knowledge, and quotes its division of animals into fourteen classes. Among them: those belonging to the Emperor, embalmed ones, suckling pigs, mermaids, stray dogs, those drawn with a very fine camel-hair brush, those that resemble flies from a distance, and those included in this classification.

Foucault put it on the first page of Les mots et les choses and said he laughed out loud.

Why is it funny? Not because the animals are strange. Mermaids are no odder than a platypus. It is funny because every category sorts by a different question. Who owns it. What state it is in. What it looks like from far away. Whether it is on this list. Linnaeus counted stamens every single time.

So you have a test, and you will use it all semester. Take any category and ask what question it answers. If two categories in one system answer different kinds of question, the system has a level problem.

What the book already sorts by#

Last thing. The Classic of Mountains and Seas is usually called a chaotic catalogue of monsters. It is not chaotic, and you cannot improve on a system until you can say what the existing one does.

Start with the cover. 山 is mountain, 海 is sea, 經 is classic, and the book divides exactly where its name divides: five scrolls of mountains, thirteen of seas. The top level of the classification is printed on the title page.

Almost every chapter title after that is a compass direction: 南 south, 西 west, 北 north, 東 east, 中 centre. Five 山經 (shānjīng) hold twenty-six numbered routes between them. Four 海外經 (hǎiwàijīng) cover what lies beyond the seas and four 海內經 (hǎinèijīng) what lies within, each set running south, west, north, east. Four 大荒經 (dàhuāngjīng) cover the great wilderness, running east, south, west, north. The eighteenth is headed 海內經 again with no direction at all.

Inside a route, each creature is filed under the mountain or water where it is found. Never under anything about its body.

So what question does this system answer? What will I meet if I go there. And it answers that completely. By the test above, that is a classification: one character, applied to every entry, never changing question halfway. Which is more than the Celestial Emporium managed.

Here is what it costs. 狌狌, 鹿蜀 (lùshǔ), 猼訑 (bóshī) and 九尾狐 (jiǔwěihú) all stand on the first route, 南山經之首 (nánshānjīng zhī shǒu), on four mountains in the order you would walk past them. Two of them have nine tails. 猼訑 has them on a sheep:

有獸焉，其狀如羊，九尾四耳，其目在背，其名曰猼訑，佩之不畏。

There is a beast there, shaped like a sheep, with nine tails and four ears, and its eyes on its back. It is called 猼訑. Wear it and you will not be afraid.

The beast four mountains further on has them on a fox:

有獸焉，其狀如狐而九尾，其音如嬰兒，能食人；食者不蠱。

There is a beast there, shaped like a fox but with nine tails, and its cry is like an infant’s. It can eat people. Eat it and you will not be harmed by 蠱 (gǔ), poisoning by witchcraft.

The text never mentions that they share a count. A system sorting by body would have put them side by side. This one files 猼訑 next to whatever else stands on 基山. That disagreement is the course.

What you leave with#

Names for the slots in a record, argued in the room. Week 2 formalises them and the rest of the semester keeps them.

And one rule that never bends, starting now: cite the line, or declare which line you could not find. A plausible-sounding quotation with nothing behind it is the single thing that fails an assessment outright. We are building a system for a text nobody can re-examine, so the citation is all the evidence there is.