Durkheim and Mauss concern themselves with symbolic classifications of a moral or religious nature, which they distinguish from practical schemes of distinctions which they call technological. They believe that the human mind lacks the innate capacity to construct complex systems of classification such as every society possesses, and which are cultural products not to be found in nature, and they therefore ask what could have served as the model for such arrangements of ideas. Their answer is that the model is society itself. The first logical categories were social categories, they maintain, the first classes of things were classes of men; not only the external form of classes, but also the relations uniting them to each other, are of social origin; and if the totality of things is conceived as a single system, this is because society itself is seen in the same way, so that logical hierarchy is only another aspect of social hierarchy, and the unity of knowledge is nothing else than the very unity of the social collectivity extended to the universe.
What follows is a passage from the conclusions of Durkheim and Mauss's article. I'm asking you to read a bit of this for a number of reasons. That their argument is correct is not among them, and if you want a list of the fatal faults in their piece, I'd again recommend Needham's introduction. However, this study introduced the topic of classification to anthropological discussion, and the history of work on classification has been important. It is part of a larger debate about the nature (or natures) of rationality as reflected in divergent cultures. One of the more interesting pieces of that work is Levi-Strauss's La Totemism Aujourd'hui, which came out in the early 60's as a sort of preface to his more comprehensive Pensee Sauvage. Indeed, the topics of totemism and classification are closely connected, in that the pre-existing work on totemism inspired much of Durkheim's thought in this whole area. [What's meant here by "totemism" is the widespread practice of naming social groups such as clans after, eg, animals, so that the set of clans becomes the set of bear, turtle, eagle, and so on. Levi-Strauss refers to this as "the so-called totemism".] It is certainly the case that all peoples classify things into distinct categories: indeed, most words refer to classes of things or actions, rather than to uniquenesses. When you refer to "trees", you refer to a class, and a class which is distinguished lexically from other classes such as "bushes" or "herbs". In that sense, classification (or generalization, or naming, or simply the drawing of distinctions) is a fundamental symbolic capacity of the human mind and one which has been elaborately studied by comparative linguistics (and by a movement in American anthropology from the 60's referred to as "ethnoscience" or "cognitive anthropology", the study of the way natural things are classified by other peoples and in other languages). But D&M have in mind something more particular: symbolic systems of classification of things of a "moral and religious nature", as Needham put it. The most commonly known system among us might be the biological taxonomy, a hierarchical scheme which begins at the top with the class of living things; then sub-divides that into plants, animals, and viruses; then sub-divides each of those and so on down to the level of the genus, which is sub-divided into species. This is the naming system invented by Linnaeus in 1735. You'll notice that D&M have in mind just this sort of hierarchically ordered classification system modeled on social divisions: the tribe is divided into two moieties, each of which is made up of several clans, each of which includes several lineages, each of which is composed of a number of families, each of which brings together several individuals. Analogous social divisions could be based on territory rather than descent, as among us: the state is divided into provinces, which are made up of municipalities.... One of the problems with their work is that other ways of classifying are possible, and indeed are more common. Can you think of any?
classifications are therefore not singular or exceptional, having no analogy
with those employed by more civilized peoples; on the contrary, they seem
to be connected, with no break in continuity, to the first scientific classifications.
In fact, however different they may be in certain respects from the latter,
they nevertheless have all their essential characteristics. First of all,
like all sophisticated classifications, they are systems of hierarchized
notions. Things are not simply arranged by them in the form of isolated
groups, but these groups stand in fixed relationships with each other and
together form a single whole. Moreover, these systems, like those of science,
have a purely speculative purpose. Their object is not to facilitate action,
but to advance understanding, to make intelligible the relations which
exist between things. Given certain concepts which are considered to be
fundamental, the mind feels the need to connect to them the ideas which
it forms about other things. Such classifications are thus intended, above
all, to connect ideas, to unify knowledge; as such, to constitute a first
philosophy of nature. The Australian does not divide the universe between
the totems of his tribe with a view to regulating his conduct or even to
justify his practice; it is because, the idea of the totem being cardinal
for him, he is under a necessity to place everything else that he knows
in relation to it. We may therefore think that the conditions on which
these very ancient classifications depend may have played an important
part in the genesis of the classificatory function in general....
...Society was not simply a model which classificatory thought followed; it was its own divisions which served as divisions for the system of classification. The first logical categories were social categories; the first classes of things were classes of men, into which these things were integrated. It was because men were grouped, and thought of themselves in the form of groups, that in their ideas they grouped other things, and in the beginning the two modes of grouping merged to the point of being indistinct. Moieties were the first genera; clans, the first species. Things were thought to be integral parts of society, and it was their place in society which determined their place in nature....
Not only the external form of classes, but also the relations uniting them to each other, are of social origin. It is because human groups fit into one another -- the sub-clan in to the clan, the clan into the moiety, the moiety into the tribe -- that groups of things are ordered in the same way. Their regular diminution in span, from genus to species, species to variety, and so on, comes from the equally diminishing extent presented by social groups as one leaves the largest and oldest and approaches the more recent and the more derivative. And if the totality of things is conceived as a single system, this is because society itself is seen in the same way. It is a whole, or rather it is the unique whole to which everything is related. Thus logical hierarchy is only another aspect of social hierarchy, and the unity of knowledge is nothing else than the very unity of the collectivity, extended to the universe....
[They go on to argue that the logical relations between things were "represented in the form of familial connexions, or as relations of economic or political subordination" and further that "the same sentiments which are the basis of domestic, social, and other kinds of organization have been effective in this logical division of things also". Then:]
is thus states of the collective mind (ame) which give birth to
these groupings, and these states moreover are manifestly affective. There
are sentimental affinities between things as between individuals, and they
are classed according to these affinities.
We thus arrive at this conclusion: it is possible to classify other things than concepts, and otherwise than in accordance with the laws of pure understanding. For in order for it to be possible for ideas to be systematically arranged for reasons of sentiment, it is necessary that they should not be pure ideas, but that they should themselves be products of sentiment. And in fact, for those who are called primitives, a species of things is not a simple object of knowledge but corresponds above all to a certain sentimental attitude. All kinds of affective elements combine in the representation made of it. Religious emotions, notably, not only give it a special tinge, but attribute to it the most essential properties of which it is constituted. Things are above all sacred or profane, pure or impure, friends or enemies, favourable or unfavourable; i.e. their most fundamental characteristics are only expressions of the way in which they affect social sensibility. The differences and resemblances which determine the fashion in which they are grouped are more affective than intellectual. This is how it happens that things change their nature, in a way, from society to society; it is because they affect the sentiments of groups differently....