Preparing for the Honolulu Jamboree of Current American Antiquity
1 April, 2013
This week On the Reading Rest I have the past in the form of a printout of an article from the latest volume, 78:1, of American Antiquity.
Because I have a grandchild in Japan I was happy to accept an invitation to give a paper at the April SAA congress in Honolulu. Nagoya is my base between Hawaii and home. Going to SAA meant that like many others of the more than 3,000 participants, I had to become a member of SAA. Being a member, whether I like it or not, whether I can pick it out from JSTOR or not, whether I read it in the library or not, whether I get a paper copy or not, I got a message, among frequent communications, telling me that I may download American Antiquity 78:1, 2013. Because I haven’t read an article in American Antiquity since perhaps 1973 when I wrote a master thesis on concept formation in archaeology – classical genus proximum et differentia specifica versus family resemblance concepts – and needed some examples of pointless New Archeology, I couldn’t resist a look.
The cover has changed, but between front and back nothing but authors’ names and reference years are new. I probably suspected just that, but who can resist a deja vue knowing that the whole antiquarian installation is staged in April – The Past, Live in Honolulu. Today, when the material remains and contexts of iterative long-lasting cultural performances are in vogue, who wouldn’t do a participant observation of the ritual gatherings of the New American Archeologists? NAAs are not endangered, not yet indigenzed and I am not doing an ethically questionable anthropology in my own society. I may be post-colonializing them, but that is still PC.
As a preparation I read:
Arakawa, Fumiyasu; Nicholson, Christopher and Rasic, Jeff. 2013. The consequences of social processes: Aggregate populations, projectile point accumulation, and subsistence patterns in the American Southwest. American Antiquity 78:1:147-165.
Arakawa, Nicholson and Rasic (AraNiRa) have observed that as of 2005 (referring to Diamond: Guns Germs and Steel) scholars have investigated the potential causes and consequences of – the Neolithic no less (p.147, first column). What a splendid idea. This Neolithic thing it seems could have had potential consequences! Thus astonished and worried – could global warming be among the consequences? – we may proceed to contemplate the following:
[Neolithic human] groups began manufacturing and using pottery vessels, used the bow and the arrow, shifted subsistence patterns from strictly hunting and gathering to horticulture and agriculture, began domesticating animals and established sedentary villages.
First year students usually don’t get away with suggesting to their teachers that pottery, the bow and the arrow and domestication, e.g. of the dog, were emblematic of anything as belated as the Neolithic. And by the way, initially among agriculturalists the sedentary villages and the distinction between manufacturing and using pottery were of little importance. In the current American antiquity, nevertheless, this new-age-dawning perspective is creed. Having introduced us to their unique cultural context the authors turn to their archaeological mission and
argue that artifact deposits from a range of settlement sizes can inform meaningful interpretations about the consequences of social processes, such as aggregation and increases in group population density.
Rather than dubious, this proposition – artifact deposits can inform a range of meaningful interpretations – is so obvious that it can be proved, i.e., probably be proved since we mustn’t jump to conclusions, must we? Where would Archeology be if we just thought it could be meaningful?
I am thrilled: reading the article, planning to go to Honolulu, believing that two of the authors will be there, I feel surfing on the forefront of that wave of research, which roles on under my surfboard without moving forward. I shall add American Antiquity to the list of consequences brought about by the Neolithic.
True to New Archeology hoping to create a meaningful pattern composed on a few hopefully controlled variables, AraNiRa start to track down (1) large and (2) small settlements, (3) high and (4) low population density in (5) a central and (6) a peripheral area, with (7) lithic projectile points and (8) earthen utility wares as well as faunal data in which we may distinguish between (9) large and (10) small game. This is not ideal because there is a dependency between lithic projectile points and large game as well as warfare, albeit perhaps a secondary one, but very little dependency between utility ware and small game or indeed warfare. New Archeology scholastics would not have taken lightly that kind of variable dependency. Variables 7 to 10 must thus be transformed to general markers, i.e. a point:sherd ration and an artiodactyle index, which are either high or low.(Google actually returns 29.900 hits for ‘artiodactyle’ e.g. beating ‘hemiepes’ with 9.000 hits).
The construction of index and ratio is always tricky. In this case it is a specimens-of-species index (based on the Number of Identified specimens in archaeological reports – NISP). That is less informative than a Minimum Number of Individuals (MNI) index, but it cannot be helped since the archaeological records give us NISP and ours not to reason why. Actually, in order not to get completely trapped in source critical problems, already troublesome, simplification is a must. The authors therefore total all the artiodactyles (even-toed), i.e. bison, deer, sheep etc, who are large, and then they sum the lagomorphs, i.e. hares, rabbits and pikas, who are small. Indeed there are no perissodactyles — odd-toed (12.700 hits). Based on these sums they construct the Artiodactyle index (AI): AI= artiodactyle/(artiodactyle+lagomorphs).
The construction of the point:sherd ratio is more difficult because it aims at reconstruction a factual prehistoric balance. This reconstruction is complicated by the fact that ‘10 out of 72 sites’ have not yielded any projectile points, just pottery. Since there is a balance between the two artefact categories, the authors manage with a bit of Bayesian statistics to produce a Posterior Mean Value, PMV, reflecting the once existing balances between points and shards. Technically speaking there is nothing wrong with the construction of this value, but the construction doesn’t differentiate between small and large sites. There is just one mean for The Core Area sites and another for The Periphery. Primarily, Core and Periphery is a matter of population density.
This means that ‘small’ and ‘large’ sites are the only straightforward, i.e. non-aggregated, variables. ‘Small’ and ‘large’ are taken for granted. In the end of the article, when it comes to comparing the meaningful patterns related to a prehistoric reality, the comparison of the PMV is primarily between small and large sites and secondarily between the somewhat aggregated periphery and core concepts, i.e. areas defined in general geographical, topographical and demographic terms. Belonging to an area makes a site ‘peripheral’ or ‘central’. In the article, methodologically motivated simplification is driven by fuzzy dichotomization keeping the reader relatively unaware of the basic distributions of the variables on the intra-site level – projectile density distribution and utility ware density distribution being cases in point.
Although this is not promising, the authors proceed to produce diagrams mapping the PMV on one axis and the AI on the other with dots representing all sites as either small or large – differentiating between core and periphery by means of symbols. In theory the expected pattern should like this:
When we start reading ‘Results’, it soon becomes obvious that there is a discrepancy between the diagrams Figures 4 to 7 and the Tables 1 & 2. There are more sites in the tables (75) than in the diagrams and the text discussing them (72). Moreover, the points plotted in the diagrams are not always the ones defined by the numbers in the tables. There are 39 tabled sites in the Core Area not the 36 plotted in the diagrams. The distribution in Fig. 7 cannot be checked because there are too few decimals in the PMV values in tables 1 & 2. There are astonishing errors too: in Fig.4 the median PMV for the small sites in The Periphery is given as 0.03, but judging from table 1 it is 0.01 and less in line with the authors’ hypotheses. This is of little importance for the obvious conclusions, but it makes patterns that are not in line with the hypotheses more significant. Only the diagram Fig. 6 (the small sites in Periphery and Core) can be reproduced and compared with the published. Based on tables 1 & 2 this revised diagram looks like this:
Superimposed upon Fig. 6 in the article we get a discrepancy:
Most of the differences are ‘decimal differences’ (the authors use four places in the diagrams and only three in the tables), but the values for McElmo-Yellow Jacket are distorted. This means that when it comes to this densely populated area the authors cannot rely on their hypotheses to explain the deviance among their large or small sites.
If we add the theoretical categories to Fig. 6, the greatest problem with the interpretation is the many small sites dominated by large-game bones, but none the less characterized by few projectile points with which to kill the game. It so happens that the upper left corner of the diagram refutes the theory on its own assumptions. This patter becomes even more obvious when we re-scale Fig. 7 and add the large sites to the small ones to get a total picture.
When we look at all sites, there is thus an even more obvious split between those with relatively speaking many and those with few large mammals, since between AI 0.4 and 0.6 there are very few sites.
There are fewer large mammals in large sites, dots, but that doesn’t mean that the animals killed in small peripheral sites, open diamonds, or indeed any small site, any diamond, were not brought as meat or cut of meat to large sites in The Core, green,blue and red dots. ‘Numerous’ and’ few’ obviously mean different things on different sites in different environments. Parallel to the split dense group, the two left corners of the diagram, there is a spread-out peripheral western group wedged between the two main corners. In this group as in the dense groups, a growing PMV means either a minimum or maximum AI. Sites in the McElmo-Yellow Jacket (McElyja) core area, moreover, are distinctly different from Mesa Verde sites suggesting that The Core Area Concept implicated by AraNiRa is flawed. If we were able to check the topographical distribution of MaElyja sites, there is a fair chance we could find a division between core sites and peripheral sites within the region.
The hypotheses suggest a theoretical distribution and meaning that doesn’t match the observed patterns. The article obscures the methodological reasons for this discrepancy, but common sense suggests that the different ways projectile points and utility ware are produced, used, dispersed, rejected, recycled, reused, deposited etc, help to obscure the meaning of the ratio point:sherd and the diagrams Figs 4-7. Probably, measuring the density of game and domesticated species as well as pottery density – leaving out the strongly dependent projectile variable, which moreover is split between killing large game and humans — as weapons still tend to be in everyday American usage – would have been more meaningful.
All this ends up in a classical example of the shortcomings of the New-Archeology approach: (1) simplification based on fuzzy dichotomies and oppositions blurs the meaning of the variables and makes correlations and discrepancies between the hypothetic suggestions and the observed patterns difficult to comprehend – for authors as well as readers. For some reason beyond our reach, (2) the formal consistency of the New-Archeology presentation often breaks down adding to the reader’s confusion. But in the end (3) this takes away nothing from the conclusions because they are but tentative and commonplace:
The analyses provide support for the idea that increasing population density and aggregation of settlement patterns led to changes in sociopolitical organization and subsistence patterns (p. 161).
It takes a backup of c. 100 references pushing at this open door. Ten seemingly innocent persons and institutions share the doubtful honour of being thanked in the acknowledgement at the very end – And of course, we [AraNiRa] take responsibility for any and all mistakes or omissions. It didn’t go without saying.