There’s nothing like a decent Bloody Mary when you really feel like a pick-me-up. Seriously. But you should know that I’m not a purist. I like to use vegetable juice, like V-8. Trader Joe’s makes a good V-8-like mix called Garden Patch that’s way cheaper and prolly better for you. I like my BMs [stop chortling, toilet-brain] on the mild side, so I add a dash or three of Worcestershire sauce and voila! Who’d’ve thought that cocktails could be so nutritious? I feel all picked up! Which is where I wanna be. ‘Cause I’m gonna pick up where I left off a while back. So, if you stick around, prepare to be catapaulted (nay, trebucheted) into the Lower Palaeolithic of Southern Africa for the rest of this blurt.
|Wilkins and Chazan (2012)*|
Remember Kathu Pan 1? All those run-of-the-mill, plain, or garden-variety flakes that Wilkins and Chazan* have interpreted as evidence of a ‘blade industry’ at 0.5 Ma? It would be an extraordinary finding if it were true. Alas. Life as a subversive archaeologist isn’t ever simple–deciding on the veracity of such a claim is not a straightforward endeavour. I fervently wish that I could just accept an archaeological inference on the face of it. Unfortunately, in the case of Wilkins and Chazan, it’s not to be.
[I know. I know. I’ve blathered on about this previously. But in truth, I’ve only called their claim into question–I haven’t as yet produced much in the way of evidence to counter or refute their interpretations.]
The truth (or reality) of this extraordinary claim hinges on the amount of morphological variability they and their ‘MSA’-of-Africa colleagues will accept into the class of flake that they’ve chosen to call ‘blades.’ To them, a blade is a ‘detached piece’ that is at least twice as long as it is wide. Simple. No? No. No archaeologist would say that it’s that simple. And those who work on the palaeolithic of southern Africa and elsewhere know that to be the case. But that’s where they start. As would be the drill for anyone who’s in the business of studying the lithic output of modern humans, the length-to-width ratio (hereafter just L-W) is a very basic starting place. The trouble with Wilkins and Chazan and others is that they try outflank the blade purists by suggesting that a quite different set of behaviours can be looked at as analogous, if not homologous, with the blade industries of modern humans. It’s a bit of a reach, as you’ll see.
Have a look at the image up above, which illustrates some of the morphological variability inherent in the Kathu Pan 1 assemblage of ‘blades’ [ignore the core in the lower left]. Here you see flakes with convergent margins, flakes that have been retouched, flakes that are nearly square, flakes that are slivers of stone, flakes with cortex, and some really thick flakes. None are parallel sided, as is the case with the Mesolithic [the true MSA] blades shown below. None are removed from prepared cores like those in the array immediately following the one below.
It’s true that modern human groups haven’t evinced prismatic-core blade industries in all places and at all times during their tenure on Earth. However, we are fairly certain that the presence or absence of such a technique has nothing to do with the cognitive abilities of the people involved–modern human decision-making and the normal constraints on behaviour often preclude such activities. Unfortunately we can’t make a similar assumption about the deep time represented at Kathu Pan 1. There, say the authors, for perhaps the first time in history a group of hominids designed and manufactured blades according to a pattern that evokes the kind of deliberate actions that we all know exist in the modern human archaeological record as part of modern human cognitive abilities.
Because it may very well be the first time for such an activity, surely it behooves these authors to be very thorough in their description of the assemblage and in the premises that underpin their argument. That’s where I’m going today [or this week, more precisely–I’ve been working on this on and off when I’ve had the opportunity, for over a week, now, which is why you’ve heard little from my pulpit in that time]. I intend to point out just how problematic are the assumptions on which these inferences have been made, and in so doing to demonstrate that the Kathu Pan 1 blades are nothing more than a reified category.
At the outset, let me say that Wilkins and Chazan have done nothing wrong by identifying certain pieces of debitage and utilized flakes as blades when they have a L-W of 2 or greater. Any good lithic analyst dealing with a modern human assemblage would do no less. But, unlike the average analyst working on the technological basis of a lithic assemblage, Wilkins and Chazan stop there. They don’t go on to describe those elongated flakes according to morphology. And that is where their analysis veers from the conventional path. They cite others who’ve worked on the MSA, including those chaps who’ve made similar claims at Qesem Cave (see the array of so-called blades from Qesem immediately below), but the truth is that this ‘blade’ classification is profoundly at odds with that of archaeologists who deal with the products of modern humans. For one thing, classic blades almost never (if ever) include cortex, as do many of those shown below.
Wilkins and Chazan and the others have decided that it’s OK to define a blade as a flake having an L-W of 2 or greater, and to go no further.
Let me just pause here to mention that they take a simple parameter–L-W–and let it represent a conscious choice on the part of the Lower Palaeolithic hominid, and from there to represent a full-on ‘industry’ like the modern human blade industry I sketched a moment ago. Theirs is a leap of the variety known as Faith, and not an argument from empirical observation, much less well-warranted assumptions.
I hesitate to go over again what makes a blade. However, I believe it’s crucial to emphasize that how we define a blade produced as part of a ‘blade industry’ is of the utmost importance in assessing the veracity of Wilkins and Chazan’s argument.
The difference between just a blade and a blade made as part of a blade industry is that the latter is the product of a conscious effort on the part of the flintknapper to produce only just such pieces and to do so in a repetitive series, in which each removal produces a blade of more-or-less uniform morphology from a specially prepared block of stone. A blade technology–one that could be said to be the product of a modern human–produces blades that bear lateral margins that are uniformly parallel or subparallel, and which display dorsal flake scars that testify to earlier removals that are of the same morphology. Among other analyses that can be done to demonstrate the systematic difference between such blades and just flakes, there is a clear-cut statistical difference between the things that have historically been called blades and those regarded as such by the authors.
Wilkins and Chazan argue that their assemblage does belie an almost identical process. They argue that the difference lies in the orientation of sequential removals, such that the prepared cores are brick-like, and that the blades removed are removed by taking one or two flakes off one end and then doing the same from the other. In so doing these knappers producing flakes that were the shape intended by their maker, i.e. the finished artifact. Unfortunately for Wilkins and Chazan, not only is this a poor argument for a stone industry thats anchored in the same cognitive ability as that produced by modern humans, even their evidence falls short of demonstrating that part of their claim.
[As I’ve said before, the subversive kind of detailed dismantling of a fallacious archaeological claim demands that no essential verbiage be spared. This is already a long blurt, and it’ll get longer. Just so you know.]
Where to begin? I thought it would be useful to see just how distinctive the Kathu Pan 1 blades were in comparison with the non-blades. Unfortunately, as I pointed out previously, the authors provide no data to enable such a comparison, nor, evidently, do they think such a comparison is warranted. In my previous attempt to discredit their claim, I as much as accused them of dissembling in their paper. However, after I had a brief, civil, email exchange with Wilkins, I retracted my thinly veiled allegation of disingenuity. However, based on what I was told in the emails, I can say that I have more reason than ever to doubt their claim.
One of the most basic assertions that Wilkins and Chazan make is that there is an emphasis on production of flakes with and L-W of 2 or greater, which they call blades. As evidence, they point their readers to their Table 2, which I’ve excerpted below.
The way this table represents the assemblage underscores my discomfort. Notice first of all that the number of what are called ‘Complete flakes’ is about the same as that of ‘Blades.’ The authors here represent that blades make up 16.1% of the entire assemblage. That is more or less the same proportion of the ‘Complete flakes.’ This implies that there are as many blades as flakes in the assemblage.
However, while it isn’t clearly stated in the table, it is mentioned in passing in the text, that the category ‘blades’ in the table above includes blades and blade fragments, while the category ‘complete flakes’ is just that. Excluding the flake fragments makes the proportion of ‘flakes’ to ‘blades’ appear almost equal. What happens if you lump ‘Flake fragments,’ ‘Proximal flake fragments,’ AND ‘Complete flakes’ and compare that number with the number of blades and blade fragments? The proportion of flakes to blades is much different, which the authors also report in their paper. They write that the proportion of blades in the total of just the ‘discarded detached pieces (including flakes and flake fragments)’ (N=3786) is 27%. This is now looking more like an assemblage in which blades are not so much ’emphasized.’
From this point their presentation becomes more confusing and disconcerting. We’re told that the mean length for complete blades is 70 mm. Then we’re told that there’s little value in presenting a breakdown of blade length, because
A frequency histogram of blade width … [see below]… can be used to get a sense of size distribution… , providing a larger sample than length because blades often break transversely
[which I guess means that we can’t know how long they were before they broke–althought it’s an open question if they were ever whole, the vicissitudes of breaking stone being what they are].
KP1 blade size distribution includes blades that would technically be classified as bladelets (<12 mm in width, … but these small blades are just at the lower end of a unimodal blade size continuum. The mean length to width ratio of the KP1 blades is 2.5:1 (n = 92, sd = 0.4).
A cartoon balloon emerges from the vicinity of my head at this point, and all it contains is a very big question mark flashing on and off like a neon sign. The number of blades in the first table above was 972. Yet, in the quote above, we’re given an average L-W ratio for blades–2.5:1 (s.d. 0.4)–based on 92, not 972. This is telling us that there are in total just 92 complete ‘blades.’ In the paper we have only one clue as to the answer–Table 4 gives the summary statistics for blades from two excavation units.
Thus, it is probable that the number 92, given as the source of the average blade length, is correct. In the subsample described in the table above, the number of blades for which an overall length was in evidence is 113. So we would be forgiven if we settled on ‘about 100’ as the number of complete blades in this assemblage.
I’m not suggesting that you need the rest of the ‘blade’ to argue that a flake portion is a portion of a blade if that flake portion is 2 or more times longer than it is wide. However, given the altogether un-blade-like morphology of the so-called blades shown above from Kathu Pan and from Qesem Cave it begs the question whether the rest of these allegedly fragmentary blades were at all blade-like throughout their length.
But, forge on we must. In the absence of (to their way of thinking) a relevant sample of complete blades from which to construct a frequency distribution of L-W, we’re given instead a histogram of blade width (shown below). Keep in mind that these widths are absolute measures of a subsample of those flakes deemed to be blades, and not a frequency distribution of the widths of the entire assemblage of ‘detached pieces.’
And when I think about it, I’m not sure what this histogram is really meant to tell us, as width only has interpretive value in this context in terms of its relationship to the length of the so-called blades. Here the number of blades is 511, because the sample that gave this result is only from 2 of the 4 excavation squares that form the entire assemblage. The authors give the average width of this subsample as 28.2 mm (s.d. 9.2) in their Table 4. In that same table the average length is given as 69.7 mm (s.d. 19.3). The histogram below clearly demonstrates that in this assemblage there is a central tendency evident in ‘blade’ width. That’s great! Except it’s next to useless unless we can see what the overall assemblage looks like on this parameter. I’d be very surprised if the just flakes produced a distribution much different from this–after all, we see variation from less than 10 mm all the way up to 60+. I doubt very much if the just flakes width distribution could look any different!
So, you and I want to know if the authors’ blades were in any measurable way distinctive from just plain old flakes. Since the Kathu Pan 1 ‘blades’ clearly don’t exhibit a uniform morphology beyond the L-W of >2, it wasn’t clear to me why we weren’t offered a similar set of data for the un-blade-like flakes. So I wrote to Wilkins:
I’m very interested in the length-to-width ratios that you reported, and I was wondering if it would be possible to acquire the raw data for the 1800 or so ‘complete flakes’ and ‘blades.’ You published the mean ratio for the blades, but not for the flakes, and you didn’t publish the frequency distribution of length-to-width ratios for either the blades or the complete flakes. Moreover, while you did publish the frequency distribution for blade widths, you didn’t publish the frequency distribution of their lengths, and you published neither for the ‘complete flakes.’
From the reader’s standpoint, on the basis of your published observations, it’s an open question as to whether or not the distincitveness of your ‘blades’ isn’t simply an artifact of the arbitrary definition of a blade.
Wilkins was kind enough to glean her data to give me a histogram of the ‘complete flake’ and ‘complete blade’ L-W ratios (N=920). It’s given below. And it’s a far cry from the sense that one gets from the histogram above.
The first thing I noticed when I saw the L-W ratio distribution for the assemblage as a whole was that complete flakes with a L-W ratio >2 (i.e. what the authors call ‘blades’) are a minority compared to those <2, and that in no way do they display the unimodal distribution that the blade width histogram above did. As I look at this histogram I don’t see a unimodal, bimodal, or a normal distribution. I see that about 700 of the 920 are ‘flakes’ (i.e. with a L-W of between 0.8:1 and 2.0:1). That works out to about a .76 probability that any piece of rock that was detached from a core and discarded in the excavated portion of Kathu Pan 1 would have had a length to width ratio of between 0.8:1 and 2:1! More fascinating, when you remember the authors’ conclusions, is that the distribution of flakes with L-W between 0.8 and 2:1 is close to that of a continuous uniform distribution. A uniform distribution occurs when one is tracking a variate that is varying randomly, as would be the case if one was sampling from a continuous variable (with replacement), and not constrained to any discrete value, such as would be the case in a game of dice. This Kathu Pan 1 distribution of the L-W ratios of all complete ‘detached pieces’ is not perfectly uniform, to be sure. But when one can predict an outcome inside 25% of the range 75% of the time, we’re talking anything but ‘normal.’
As well as being platykurtic this distribution is heavily skewed to the lower L-Ws. As for those flakes that Wilkins and Chazan would call blades, the numbers taper off to the right much as would be the case in any platykurtic normal distribution. This is to be expected even in a distribution that is close to uniform for a portion of the range. You simply wouldn’t expect a strongly, but imperfectly random process to suddenly drop to zero at any point. Thus, the shape of the distribution for the longer flakes is what you’d expect, even from a process that was more random than not.
I believe that the L-W ratios of complete blades and flakes data more or less destroy the authors’ contention that ‘blades’ were preferentially removed, relative to just plain old L-W <2 flakes. There's nothing distinctive whatsoever about their blade dimensions when compared along with the rest of the assemblage. I fully expect a similar outcome if we were ever to see the frequency distribution of all 'detached pieces,' both fragmentary and complete, whether called 'flakes' or 'blades' by Wilkins and Chazan.
Alas, I wish I could say that we’re finished. We still have to deal with the authors’ contention that there are these prepared cores and that the ‘blades’ frequently demonstrate ‘bi-directional’ flake removals, which the authors would say was analogous to the prismatic cores that we see produced by modern humans.
When I asked as to what it was besides the L-W ratio that distinguishes the flakes they call blades as blades, Wilkins replied
the majority of the detached pieces that are twice as long as they are wide have bidirectional dorsal scars (relating them to the bidirectional cores that we describe)
Let’s look closer at their contention. I don’t know any other way than to display what the authors have given us, at a scale that makes sense. These views of the so-called bi-directional cores are more or less actual size. After admiring the lovely layered effect of banded ironstone in the photo, have a look at the drawing labelled ‘c.’
|A ‘bi-directional core’|
But that’s still not all. The authors aver that the flakes they call blades show dorsal flake scarring indicative of at least one previous removal going in the opposite direction–i.e. evidence of having been struck off one of their fantasy bi-directional cores. They say that even the complete flakes don’t show the same dorsal morphology. Here’s what I think.
And if you don’t believe me about the brick-shaped core thing, just listen to what Wilkins and Chazan say about their wonderful, prepared, bi-directional cores.
The authors use the term idealized a lot in this illustration. Go figure! Here they provide no evidence of the presumed preparation of a chunk of rock to produce such a shape from which to strike off elongated flakes bi-directionally. And even if you give the hominids the benefit of the doubt (the half-million year old ones, not the authors), all we’re seeing here is, perhaps, the result of a choice as to which of the four ‘sides’ to exploit.
Of course, what I think doesn’t matter a hill of beans. The palaeoanthropological ‘community’ will think whatever they want to think, regardless whether the evidence is real or trumped up. So, I’m gonna go back to the store now and get me some more vegetable juice!
* Wilkins, J., Chazan, M., Blade production ~500 thousand years ago at Kathu Pan 1, South Africa: support for a multiple origins hypothesis for early Middle Pleistocene blade technologies, Journal of Archaeological Science (2012), doi:10.1016/ j.jas.2012.01.031
Please show your support for this blog. Remember that when you purchase from Amazon by clicking any of the links at this blog, you’ll be getting great discounts and helping the Subversive Archaeologist to pay for his fieldwork at the same time!