The streetlight effect, yet again: The way Scythian origins should not be approached A.V. Gromov, A.A. Kazarnitsky, A.G. Kozintsev Museum of Anthropology and Ethnography, Saint-Petersburg, Russia Our note is prompted by the article Nonmetric cranial trait variation and the origins of the Scythians by A.A. Movsesian and V.Yu. Bakholdina (hereafter M&B), recently published in the American Journal of Physical Anthropology (Movsesian, Bakholdina, 2017). The objective of their study was to use the multivariate analysis of frequencies of 32 nonmetric cranial traits (hereafter NMTs) to test the two competing scenarios of Scythian origins––autochthonous (originating from people of the Late Bronze Age Timber-Frame, or Srubnaya culture) versus resulting from a migration from Eastern Central Asia. It would be wrong to claim that M&B have failed to accomplish their task; rather, they didn’t begin to approach it. This might be partly due to their expectations regarding the potential readership. Apparently, the first thing that captured the attention of the editors, the reviewers, and many Western readers alike was not the gist of the matter (with which most of them are less than intimately familiar) but the formal aspect–– statistical analysis, formulae, plots, English writing style, etc. All this M&B have demonstrated brilliantly, and the outcome is an article which is formal in the most literal sense. We can hardly imagine such a publication in a Russian journal destined for specialists in population history and archaeology. To begin with, their database is entirely unsuited for the purpose of the study for several reasons. All the five Scythian series are quite late, dating to late 3rd century BC – 3rd century CE. We are at a loss to find the reason for such a choice. Early Scythian samples are admittedly very scarce, but those dating to the 5th–4th centuries BC are abundant. Regarding the chronological gap between them and the late Scythians as insignificant testifies to being unversed in the issues raised. Numerous events occurred in the population history of the region during those centuries. Specifically, the influx of Asian groups increased, and separating this late genetic signal from the early one, supposedly related to the appearance of Scythians on the historical scene, is impossible with such a database. M&B do not appear to have cared either about the dates or about the ethnic attribution of groups. For instance, the sample from Zolotoye is attributed to late Scythians in the article and no further details are provided. However, the excavator of this cemetery, V.N. Korpusova, believed that those buried there were Bosporan Greeks (Korpusova, 1983, pp. 31, 75, 96, 97). T.S. Konduktorova, who had measured the crania from Zolotoye, stated that those people were mostly Greeks and Tauri, and that the available data “indicate neither the absence of Scythians in the population nor their considerable share therein” (ibid., p. 174). The bearing of the Sarmatian samples, which are contemporaneous with late Scythian ones, on the question of Scythian origins is hard to grasp. As to Bronze Age groups, Eastern Central Asia is represented by two samples from the Minusinsk Basin––Andronovo and Okunevo from Chernovaya VIII. None of them has ever been mentioned by archaeologists or physical anthropologists as supposedly ancestral to the Scythians. The choice of the former group is simply unmotivated whereas the use of the latter stems from a misunderstanding. M&B cite the study by Kozintsev (2007), who, in their words, wrote about the craniometric resemblance between the Scythians and the Okunevo people of southern Siberia. But the notion “southern Siberia” is much too vague. What one of us referred to was solely a small Bronze Age sample from Aimyrlyg in Tuva. Its size precludes its use in nonmetric analysis. Craniometrically, this sample is very different from the Okunevo groups proper and is indeed strikingly similar to the steppe Scythian populations. However, it has no direct relationship to the Minusinsk Okunevo populations either in terms of craniometry or in terms of archaeology. Certain archaeologists, primarily V.A. Semenov (1997), did use the term “Okunevo culture” with reference to Bronze Age sites in Tuva because the ceramics found there resemble those of Okunevo. However, such ceramics are associated with other cultures as well, for instance with Samus (Esin, 2002), which is similar to Okunevo in many respects (Savinov, 1997). E.U. Stambulnik and K.V. Chugunov (2006), who had published the Bronze Age finds from Aimyrlyg, prefer the term “Okunevo-like culture” while stressing its distinctiveness and the fact that it “had its own destiny, different from that of the Okunevo tribes in the Minusinsk Basin.” Most archaeologists speak of “Okunevo-like cultures,” the “Okunevo cultural tradition,” the “Okunevo milieu,” or the “Okunevo-type group of cultures,” which, apart from Okunevo proper, includes Samus, Karakol, Krotovo, Kanay-type burials in Kazakhstan (see Savinov, 1997 for a review), and finally the Okunevo-type culture in Tuva, for which V.V. Bobrov (1992, 1994) proposed the term “Chaakhol,” thereby eliminating the confusion. M&B, however, spared themselves the trouble of analyzing these intricacies; in fact, they preferred to ignore them altogether. Their logic is quite plain: if the Chaakhol sample is too small for nonmetric analysis, why not use the large Okunevo sample from the Minusinsk Basin as a proxy? This illustrates the well-known streetlight effect (the story goes that a man who has lost his keys in the park searches for them under a streetlight because “this is where the light is.”) Unlike him, however, M&B are trying to convince their readers that they are searching in the right place. It does sometimes happen that a misunderstanding helps us hit the right road by a lucky chance. But the hypothesis M&B are testing is not merely nonexistent––it is utterly implausible even in theory. Craniometrically, the Okunevo people of Minusinsk with their brachycrany, flattened face and other distinctive features (Gromov, 1997) are infinitely remote from the Scythians. As we demonstrated, they can be regarded as collateral relatives of the American Indians (Kozintsev, Gromov, Moiseyev, 1999). Our conclusion, based on the integration of measurements and informative NMTs, has been supported by recent paleogenetic studies (Allentoft et al, 2015). It is hard to find less suitable candidates for ancestors of the Scythians among the Bronze Age groups of Eastern Central Asia. The Mahalanobis D2 distance corrected for sample size between the steppe Scythian group of the 5th – early 3d centuries BC and the Minusinsk Okunevo sample is tremendous – 13.44, whereas that between the same Scythian series and that of Chaakhol is negative (–0.23), i.e., less than the statistical error. In other words, those samples are virtually identical. But maybe M&B have demonstrated a nontrivial similarity between the Scythians and the Okunevo people of Minusinsk in some exceptionally informative NMTs? Nothing of the kind! What they have actually demonstrated is the utter inutility of their battery of randomly selected traits for the solution of the problem they are trying to resolve. The advantages of carefully selected and informative NMTs over randomly picked and mostly useless ones have been discussed more than once (Kozintsev, 1988, pp. 5-8; 1992). A single example will suffice here: the Minusinsk Okunevo sample in M&B’s analysis turns out to be even closer to the Srubnaya group than to any Scythian series. The MMD distance in this case is negative, implying that the two samples are indistinguishable! A more absurd result can hardly be imagined. Will M&B seriously claim to have demonstrated a close relationship between the Srubnaya population and the Okunevo people of Minusinsk? We will illustrate all the above by comparing two plots––that from M&B’s article (Fig. 1, left) and that based on the craniometric analysis of the same groups (Fig. 1, right). In the latter case, D2 distances were processed by the same NMDSCAL algorithm in the STATISTICA 12.0 package. Let the readers themselves judge which of the two plots shows a better agreement with archaeological and geographic data. Fig. 1. The arrangement of samples based on NMTs (left) and on craniometric traits (right). Can any meaningful information be gained from the left plot? It can, but the information is rather trivial––it mostly relates to chronology (dimension 1): the more ancient a group is, the further it is from the late Scythians on average. Accordingly, the proximity of the Scythians to the Sarmatians indicates not so much their relationship as the fact that the former differed from their neighbors and contemporaries less than from people who had lived millennia before their appearance on the historical scene. The same chronological factor evidently accounts for the intermediate position of Srubnaya and Minusinsk Okunevo people and for their resemblance, in fact, inseparability, for which no rational explanation except the chronological can be found. Being faced with the same situation, we extracted the chronological vector from the total variation and excluded it from further calculations because no genetically relevant information could be gained from it (Kozintsev, Gromov, Moiseyev, 1999). In conclusion, we reiterate an age-old rule: a formal approach to a problem inevitably yields a likewise formal and uninformative result. Bibliography Allentoft M.E., Sikora M., Sjögren K.-G. et al. Population genomics of Bronze Age Eurasia – Nature, Vol. 522, no.7555, 2015, pp. 167-172. Bobrov V.V. The Kuznetsk-Salair region during the Bronze Age. Novosibirsk, 1992 (in Russian). Bobrov V.V. On the migration of Caucasoids to southern Siberia during the Seima period. In: Paleodemography and migration processes in western Siberia in prehistory and during the Middle Ages. Barnaul, 1994, pp. 53-56 (in Russian). Gromov A.V. The origin and relationships of the Okunevo population. In: Okunevo collection. Saint-Petersburg, 1997, pp. 301-345 (in Russian). Esin Y.N. On the relationship between Samus and Okunevo cultures. In: Proceedings of the V.M. Florinsky Museum of Archaeology and Ethnography, vol. 1. Tomsk, 2002, pp. 39-56 (in Russian). Kozintsev A.G. Ethnic cranioscopy. Racial variation of cranial sutures in modern man. Leningrad, 1988 (in Russian). Kozintsev A.G. Ethnic epigenetics: A new approach. – Homo, vol. 43, no.3, 1992, pp. 213-244. Kozintsev A.G. Scythians of the North Pontic region: Between-group cranial variation, affinities and origins. – Archaeology, Ethnology, and Anthropology of Eurasia, vol. 4 (32), 2007, pp. 143-157. Kozintsev A.G., Gromov A.V., Moiseyev V.G. Collateral relatives of American Indians among the Bronze Age populations of Siberia? – American Journal of Physical Anthropology, vol. 108, no.2, 1999, pp. 193-204. Korpusova V.N. The necropolis of Zolotoye (on the ethno-cultural history of the European Bosporus). Kiev, 1983 (in Russian). Savinov D.G. Issues in the study of the Okunevo culture: the scholarship. In: Okunevo collection. Saint-Petersburg, 1997, pp. 7-18 (in Russian). Semenov V.A. The Okunevo sites in Tuva and the Minusinsk Basin: comparative characteristics and chronology. In: Okunevo collection. Saint-Petersburg, 1997, pp. 152-160 (in Russian). Stambulnik E.U., Chugunov K.V. Bronze Age burials at the Aimyrlyg burial ground. In: Okunevo collection 2. Saint-Petersburg, 2006, pp. 292-302 (in Russian). Movsesian A.A., Bakholdina V.Yu. Nonmetric cranial trait variation and the origins of the Scythians. – American Journal of Physical Anthropology, vol. 162, no.3, 2017, pp. 589–599.