Flat files

[Mayline - Stacor Cole, Safco Flat File Cabinet 5 Drawers for Drawings, Maps, Blueprints (Serviced) - In 10 baskets - Price: £463.59+. Turned up by Bing on the clue ‘flat file’]

A bit of tutorial material associated with my ongoing attempt to fathom Lévi-Strauss’s use of simple diagrams and mathematical jargon at reference 1.

When I was young, I had a lot to do with what we then called flat files. A flat file was a series of records of fixed length and fixed format: all the records were about the same sort of thing and had the same format, that is to say they were organised in the same way. Very much analogous to the form that you might fill in for some bureaucratic purpose. An example might have been records about deaths in England and Wales during 1974. Such files were often large and held on magnetic tape – and a magnetic tape reader was then the size of a walk-in refrigerator.

To be distinguished from hierarchical files, often a two level hierarchy, for example persons within households. One would have one sort of record about households, another sort of record about persons. From an analysis or a computing point of view, a much more complicated object to deal with.

While a lot of the data in a flat file was properties like age (numeric) or sex (categorical), the rules also allowed for count fields like number of pets or number of stays in hospital in the last ten years. Such counts would mitigate the absence of a hierarchical structure which would better accommodate said pets and visits.

Such flat files were rather like the Excel worksheets which came some years later, although, at least in the beginning, a flat file could manage a lot more records – perhaps hundreds of thousands – than Excel could manage rows.

An example

 This example of a worksheet being extracted from reference 2, with the rows being all the different species, sub-species and hybrids of oak tree, more than 800 of them. A classification; much more the sort of thing that Lévi-Strauss would take an interest in.

Note that while the display is fixed format, some of the fields contain variable length text. A flat file of old would have to fix a maximum, which while simple to process, can be expensive in (white) space.

I would fault the worksheet in using more the one header row at the beginning. There are also some footnotes at the end. One can see the need, but Excel works much better when there is at most one header row and nothing tagged on at the end.

One possibility with this type of file is sorting. If one has columns corresponding to the identifying properties of, in this case, oak trees, one can sort on those columns to give something which is not that far removed from a hierarchical tree structure, the sort of tree structure that a taxonomist might come up with. Or the binary key for identification that a botanist might come up with.

As it happens, the file shown is sorted alphabetically by name and there are not enough of the sort of properties needed to provide useful structure – but one can cheat by choosing one of the various taxonomies which are provided.

The project

One can then view the Lévi-Strauss project which is the subject of the paper at reference 4 as an attempt to come up with enough properties to enable comparative analysis of all the cooking vocabularies around the world.

Where properties are preferably binary, but otherwise mostly categorical with a small number of values. He also allows:

• A property to be absent. Some properties are only relevant in parts of the vocabulary

• The value of a property to be unknown or not stated

• The value of a property to be qualified. 

So one might have a property called heat, which is present or absent, with the optional qualifier being the source of the heat: charcoal, wood, oil, gas or whatever. I duck here the debate about whether one can be cooking if one is not using heat.

Binary properties might be mapped onto good or bad: that is to say one value of the property is mapped onto good, the other onto bad. Although such mapping might depend on context: if one is overheated, hot is bad – while if one is very cold, heat is good.

Any given cooking word can then be described, in large part, by the values it takes for those properties.

So we might have it that: fry╞ cook(heat=present fat=present water=absent container=present(frying pan)).

Avunculate

This came up at reference 3 and is described more generally reference 4.

The snap above being taken from a worksheet version of the avunculate figure there, a worksheet which lists all the permutations of the properties available. With ‘informal’ and ‘formal’ being a shorthand for the usual relationship between the pair identified in the corresponding column heading – aka the meta-data – of the worksheet.

The figure at reference 3 is not exhaustive, but the hypothesis is that all the permutations which are not excluded do in fact occur somewhere in the world – albeit some with greater frequency than others.

The point being that this worksheet provides support for systematic investigation of the worldly variation in such matters. It is not the particular permutations which are important in themselves, rather the choice of properties (and their values) defining a particular group of customs. A group comparable to the ‘semantic field’ – exemplified by the word ‘cooking’ – used by Lehrer at reference 5.

One can examine all the permutations and see which ones actually crop up – and why.

One can examine the extent that the hierarchies implicit in various sort orders correspond to the corresponding field vocabularies of the peoples of interest. To the things that anthropologists think important.

The diagrams, such as the culinary triangle of reference 5, are not much more than a bit of trim.

Something which data analysts in IT know all about when they design the relational tables in the (relational) database supporting an application on a computer – with the relational table being another variety of the flat file, albeit a rather powerful – and now ubiquitous – one. What SQL is all about.

Conclusions

We suppose that human brains are pretty much the same everywhere. We have not been around for long enough to there to have been much divergence.

The basic drives to survive and to reproduce are the same everywhere: the drives for food, drink, shelter and so forth. Maybe you get the Freudian ego, id and superego and their derivatives everywhere.

We suppose that the facts of cooking physics are pretty much the same everywhere.

However, the environment and context of cooking vary a good deal. For example, the sort of food that is available and the sort of cooking technology that is available. Not all people, for example, have pots (for boiling) – something that Lévi-Strauss himself draws attention to in the short paper at reference 6 – rather complicating his focus on boiling food.

Language varies a good deal too, although linguists have worked hard to uncover universals, under the hood, as it were.

All of which means to me that the supposed Lévi-Strauss hypothesis that there is a set of cooking properties which are both efficient and sufficient for the analysis of the cooking vocabularies of the world is interesting. It would be interesting if this were so, but it is not obvious (to me anyway) that it is so.

And where such a set of properties would sit in the chain of causation is another matter. To what extent would we be conscious of those properties and able to articulate them to a skilled investigator?

Exclusions

The foregoing does not address the business of mappings and transformations more generally, more mathematical flavourings, also important in the world of Lévi-Strauss.

References

Reference 1: The Jealous Potter – Lévi-Strauss – 1985, 1988.

Reference 2: International Oaks - The Journal of the International Oak Society – 2018. No.29. Contains 'Updated Classification of Oaks: a Summary – Béatrice Chassé – 2018. Floraoddmenu_829, 830a, 830b.

Reference 3: https://psmv6.blogspot.com/2026/10/wilcken-on-levi-strauss.html. 

Reference 4: https://en.wikipedia.org/wiki/Avunculate. Plenty of Lévi-Strauss here!

Reference 5: Cooking Vocabularies and the Culinary Triangle of Lévi-Strauss – Adrienne Lehrer – 1972.

Reference 6: Le Triangle Culinaire – Lévi-Strauss – 1965.