m UNIVERSITY LIBRARY ^: miEmiY OF CALIFORNIA, SAN Diu 0=: ->^«A iClM, CALIFORfm MIGRATIONS. (The Huxley Lecture for 1906.) BY Peof. W. M. FLINDEBS PETEIE, D.C.L., F.R.S. [WITH PLATES XIX-XXIX.] \_Reprinied from the Journal of the Aiithropological Institute, Vol. XXXVI.'] PUBLISHED BY THE ^nf9to|>ofo5tcaf Jngftfufe of (great fgxiUxn nnb Jrcfanb, 3, HANOVEK SQUAEE, LONDON, W. LIBRARY UNIVERSITY OF CALIFORNIA SAN DIEGO J MIGRATIONS. {The Huxley Lecture for 1906.) By Pkofessok W. M. Flinders Petrie, D.C.L., F.K.S. [Presented November 1st, 1906. With Plates XIX-XXVI.] The growth and decay of races, their changes and movements, form a large part of the study of man. To trace these features in the past is one of the principal results sought amid the enormous mass of measurements and details which are being recorded. But amid this mass of material, which is so vast, and yet such a small fraction of what is needed to comprehend the facts, we require some orderly progression, some systematic precedent for its ^interpretation. In this, as in all research, we must proceed from the known to the unknown, and the historical records of races must be our guide in learning how to interpret their remains. We must learn the methods and grammar of the physical anthropology by its relation to historical facts. Such seems to be the tirst need of this science at present ; and all that I can hope to do in this lecture is : (1) to give an outline of the general considerations bearing on the mutations and movements of races ; (2) to sketch the racial history of one country that we know best historically — Egypt; and (3) to give an outline of the changes in one great period where they are best recorded, the convulsions of Europe from Augustus to Charles the Great. Each of these subjects might well occupy a long course of lectures ; and I am painfully aware of the fragmentary nature of what I can now offer in a single paper. But the absence of any modern work dealing with these subjects from an anthropological standpoint, must be the sufficient ground for tlic present endeavour to stimulate students to produce some adequate researches on these important enquiries. In dealing with questions so far-reaching, and at the same time so full of minute detail, our scope must be severely limited ; the information so important to the historian about the personage and the exact places of the movements must be entirely set aside ; and the great political changes of governance, the brilliant raids and the forlorn hopes, have nothing to do with our subject. Our purpose should be to compare and understand the racial movements where they come into historical view, so that they may show the true interpretation of those physical changes which are our sole informants concerning most of the past of mankind. [2] Peof. W. M. Flinders Petrie. — Migrations. ISIigration may be described as an animal habit, whether we regard temporary migration, to and fro with the seasons, which the birds perform on the largest scale ; or the permanent occupation of new ground, which has been the necessary progress in the growth of every species. The great seasonal march of the bison up and down Xorth America, and the consequent movement of its hunters, was the type of the movements which doubtless existed in Europe, and from which grew the annual migration of pastoral nomads in search of summer and winter pastures, A curious trace of such movements may be seen in Yorkshire, at Scamridge Dykes, between the head of a branch of the vale of Pickering, and the head of another valley system. There a narrow neck of land is deeply scored with dozens of cattle-tracks, some as much as five feet deep, and all gradually worn by the passage of herds of cattle from one valley to the other during long ages. The permanent occupation of new ground has never been on a greater scale than it is at present. The ease of movement now has led to half the world being in course of occupation by the other half. Eaces are exterminated, and wholesale changes are going on now in a lifetime w^hich might have occupied a thousand years in past ages. And not only is there migration, but there is also mixture which always accompanies it. The multitudes of Eurasians in India, the whole tribes of Dutch-Bushmen in Africa, the Xegro-Americans, the complexities of South American-Xegro-European mixtures, the Dutch-Javaus, the Scottish- Canadians, all show how ine\dtably fusion of even most diverse races will occur. The common verdict that mixed races inherit the vices of both and the virtues of neither parent, is only due to the unhappy fact that the parents of such mixtures are usually those who have far more vices than virtues to transmit. The noble ideal of Alexander at the great marriage of the East and West in the Babylonian plain, when Greek and Persian promised to unite in one great ruling race, has seldom been carried out. But many splendid examples of diverse parentage show that the causes of failure lie rather in character than in diversity. In our own country also we may see the progress of gradual migration, the slow continuance of those great movements which have produced the present population. Those who question the continuance of British parentage among the Saxoi» population have taken little account of the universal fact that women of a conquered race are always incorporated with the conquerors, " to every man a damsel or two " ; and also they have neglected the back-flow of the British during peace. A striking recent instance of Welsh penetration of England may be seen at Hereford, where there is scarcely a Welsh name inside the city, while a new road between the city and the station is mainly occupied by Welsh. Even in the East of England the mixture has been noticeable. In the upper classes of Sussex there are 58 per 1,000 of Welsh names, and nearly as large a proportion in London. The examination of names of the upper classes shows that in Sussex there are 20 per cent. un-English : in a rather foreign quarter of London, the north- west, there are 30 per cent. un-English ; while of Sussex farmers there are but 2 per cent. un-English. Thof. W. M. Flixdeks Petkik. — Miijrations. The details are, per 1,000 : — [3] Loudon N.W. Sussex, private. Sussex farmers. English ... 698 803 978 Welsh 44 58 7 Scottish... 86 65 11 Irish 10 25 2 French ... 21 23 1 German... 101 26 Others 40 (In French.) 1 These represent the mixture since the introduction of surnames. Probably considerable mixture had also taken place before that period. Thus we see that the Saxon occupation is being largely changed by a more eastern migration of semi-Slavonian Germans. The results of migration are very different according to the nature of the changes which it imposes. The forms may be classed thus : — Displace- ment by occupying emptied lands, ,, pushing on another people, „ entire massacre, Tof men only, I ,, women only, Mixture ■{ I „ men and women. as Lombards in Po valley. „ the Huns pushed the Goths. „ the Tasmanians by the English. „ American-ISregro mulattoes. „ army of Claudius, which took 50,000 women. „ in German mixture in London now. The consequences of all these different forms of mixture depend upon four variables, and a main purpose of anthropology should be to attahi some estimation of these factors of change in a race. They are (1) Plasticity of race. For instance this appears to be facile in English, as a foreigner has remarked on the difference of type produced in a few generations of different conditions in the various colonies ; and it appears to be resistant in Jews, among whom peculiarities of face and expression are alike after segregation for thirty or forty generations. (2) Force of environment; which may be very slight as between similar climates, or very severe in very diverse climates. (3) The amount of mixture with another race, which may vary not only directly but also by subsequent conditions tending to eliniinate one of the races, artificially or naturally. (4) Time. In dealing with the history of migrations it appears that there is a previous movement some centuries in advance of a general migration. This usually consists of the more active men coming in as raiders or as mercenary troops, some of whom often rise to leading positions among the earlier inhabitants. This requires notice, as such forerunners are liable either to be mistaken for the main body, or to be entirely discredited because they are not the main body. We may here notice some instances, beginning with the latest. There is evidence of a 2 [4] Prof. W. M. Flinders Petrie. — Migrations. S}Tian influence, and of a body of Saracen cavalry, in the eastern side of Egvpt as early as 402 (in the Xotitia), that is, 238 years before the Arab conquest of Egypt. The well-known Saxon shore of England is another instance ; the conquest of Britain by Carausius with his Dutch pirates in 287, at the same time that the Saxons infested the Gaulish coasts, the raids of the whole coast of England by 350, the Saxon occupation of the more distant coasts of Xormandy by 370, and the Saxon shore of England, already so named, in 402, were all stages in the preparation of the great migration one to two centuries later. Tlie Huns appear about 100 A.D. north of the Crimea, and at that time Decebalus, the king of Dacia, bore the same name as the great Mongolian khan four centuries later, Dizabul — Dizabulus — as Latham pointed out. Forerunners of the Hunnish migration of 425 appear here. Earlier we notice the body of German guards of Caligula in Eome, centuries before these people conquered Eoman territory. Looking to Egypt we see the Greek mercenaries employed by 664 B.C., more than three centuries before the Greek conquest of Egypt. Similarly we read of a Hittite chief in Palestine about 1800 B.C., some five centuries before the Hittites descended from theii^ Armenian home. Before that there are remains of eastern rulers in Egypt, probably some centuries before the Hyksos invasion. And still earlier the decorated buttons of barbarian manufacture which belong to the invaders of the YII-YIIT Dynasties in Egypt, are first found a century or two earlier, showing that the foreigners were coming in during the time of the pyramid-builders. Though attempts have been made to deny, or explain away, some of these instances, the uniform nature of these examples show that we have here a general feature of migrations. We now turn to the view of the history of one country for about 10,000 years. In the absence of exact data, from even the best-known lauds, in regard to physical changes in man, Egypt may give us perhaps a better general view of historical and material changes in a people than we can reach elsewhere. Of the palaeolithic man no remains but flints have been found. But the extreme freshness of surface flints of palaeolithic type, found at the present low level of the Xile, suggests that there has been no long neolithic age as in Europe. This is borne out by the conditions of the country. While there was a few inches of rainfall in Egypt, making up for the loss of the Xile bysoakage and evaporation, there was no cause for the alluvium to be deposited. At that time no agriculture was possible, from the absence of alluvial flats and the slightuess of the rainfall. When the rain ceased the mud was deposited by the lessening river, and agriculture became possible by irrigation. In the earlier age the Xile valley would have had a deep gorge channel, with a stream rapid enough to carry all its mud; some amount of trees and wild herbs would sustain animals, and man would be a hunter. The strata at the close of this age, the beginning of alluvium, are at a depth of about 10 metres under the present surface, corresponding to about 10,000 years of deposit. This date leads us so nearly to the beginning of Prof. W, M. Flinders Petkie. — Migrations. [5] the continuous civilisation, that it seems probable that the hunting man was ejected by the agriculturist so soon as cultivation was possible iu Egypt. The late condition of the flints found point the same way. Now it is remarkable that in the earliest graves which we know, probably 9,000 years or more in age, many figures have been found with the Bushman, or Koranna type of steatopygy. And these figures are always painted red, while the figures of European type with them are white. The steatopygous type in the French caves is shown, even in females, as being hairy over the body ; and the Egyptian female figures of the same type have hair along the lower jaw. It seems that this earlier race was the same as that known in France, in Malta even in the times of temple building, later in Somaliland, and now only in the extreme south of Africa. And it may not be unreasonable to see in this the last remains of the palaeolithic man of Egypt, whom we can thus restore to view as a steatopygous and hairy Bushman. The figures found being all female, and apparently put into the graves as slave- models, agree well to their representing the captive slave woman of a disappearing type, partly expelled, partly exterminated. An entirely different people succeeded these, of European type, tall, slender, pale, with long brown wavy hair. Throughout the long age of the prehistoric civilisation we find no marked difference in the figures of this type until the beginning of the dynastic conquest. The best portraits of the type are given in the Jour. Anth. Inst., xxxi. The high, well-domed head, the long, sub-aquiline nose, and the pointed beard are constant in all the figures. Whence came these people ? They are unlike anything from the south, and the portraiture is not at all that of the Semite on the east. We are not justified in expecting any considerable water transport in such a barbaric state, so either Syrians or Libyans must be regarded as the most likely invaders. Now it is among the Amorites of Syria and the Libyans that we find exactly the same facial type (see heads in Jour. Anth. Inst., xxxi, PL XVIII), and it is agreed by all that there are very close resemblances in the Libyan culture. We may note the following connections : — (1) Profile closely like Libyan and also Amorite. (2) Colour fair, as modern Kabyle and ancient Amorite. (3) Pottery hand-made and burnished, as Kabyle. (4) ., faced with haematite, as Kabyle. (5) ,. decorated with white slip lines as Kabyle. (6) „ patterns, geometrical, as Kabyle. (7) „ cross line decoration, as on Libyan men. (8) Flint work delicate, as armlets from Sahai-a graves. (9) Tatu patterns, like those on Libyans. (10) Eoyal title bati, as Libyan hattos. (11) Crown on pottery, as crown of Libyan goddess Neit. (12) Emblem of goddess Neit on Libyan tatu. [6] Pkof. "\V. ]\I. Flinders Petrie. — Mifjrations. Now as against this, on the other hand there is — (13) Pitchamber burial, instead of dohuen burial of Libya and Syria. (This is a natural result of the conditions. In Egypt it is difficult to pick up blocks of stone, and there are gravels to dig in. In Libya the cemeteries are on tracts of rocky ground wheie blocks abound, and no pit could be dug.) (14) The prehistoric language is unknown, but the later Egyptian comprises much Semitic in structure, though most of the words are of other sources. (This only shows a Semitic mixture ; and we know that a minority of Arabs have sufficed to sulistitute an entirely Semitic language since then. In no case can language prove a race descent, as all anthropologists know.) We now turn to the evidence of skidl measurements, and these will 1)0 here treated directly as lengths., and not compounded in ratios, as it is desirable to know what elements vary. We shall first notice the facial measures, as it will be seen (in the annexed paper on the interpretation of curves), that dimensions of a single bone are preferable to those which depend on many variable qualities, as in the length and breadth. Eeferences to curves 1-79 belong to illustrations in the interpretation of curves. The median is always used here instead of arithmetical mean, as it is less liable to casualty in dealing with short series. Nasal Height. — This in the early prehistoric has a mean of 50 millimetres with probably two groups at 48 and 52i (curve 30). Later in the preliistoric age the main group is at 50, and a new small group at 53^ (curve 35). The Koknia tombs in Algeria show a mean of 50 ; but as that is mixed male and female, and female is 2 less than male in Egypt, we should put 51 as corrected to male value. The measures of living Algerians give, Chawia 51, and Kabyles ol\ for males. (See Maclver and Wilkin, Lihijan Notes.) These summarised are : Early prehistoric ... 50, groups 48 and 52i, Late „ 50, new group 53 J. Algerian tumbs ... 51. Living Chawia ... 51. „ Kabyles ... oil. With a mean variation in these of 2 millimetres either way and a range of 14, a difference of 1 millimetre is quite insignificant. The Algerian falls between the two components of the Egyptians. Nasal Breadth. — There was none measured of the Algerian tomb series, and the measures on liAiug Algerians do not refer to the bone, but to the flesh. In general terms both Egyptian and Algerians are mesorhiue, the Egyptian low mesorhine, the Algerian high mesorhine. Some admixture of a lower race in Egypt woidd make this dittereuce, which is however all within one class. Prof. W. M. Flinders Petrie. — Migrations. [7] Nasi-alveolar height. — This in early prehistoric falls in two groups with means at 66 and 73 (curve 32) ; in later prehistoric these were fused into one at 69, and a new small group arrives at 75. This measurement is not in the Algerian tomb series. The living Chawia give 71, grouping rather on 67 and 72, and the living Kabyle give 69J : Early prehistoric . . . 68^, groups 66 and 73. Late „ ... 69, new group 75. Living Chawia ... 71, groups 67 and 72, „ Kabyle ... 69|-. Here no distinction can l)e made, the early groups are almost the Chawia groups, and the later type is close to the Kabyle. The variations are insig- nificant. Bizygomatic Ircadth. — This in early time was about 125, groups being probably at 122 and 129 ; and later it was about 128, with a main group at 126, and a minor group at 133. The Algerian tombs give 127, which might perhaps rise to 130 if corrected for female example. The living Algerian is of coiirse not com- parable exactly ; the Chawia gives 136^, with groups at 131 and 138, and the Kabyle 139, Probably 7 may be deducted for the flesh ; thus we should have : Early prehistoric ... 125, groups 122 and 129. Late „ ... 128, „ 126 ., 133. Algerian tombs ... 130 ?, mixed sexes 127. Living Chawia ... 129|^, groups 124 and 131. „ Kabyle ... 132. Here the groups in the late and early Egyptian types are on either side of the Chawia groups ; and it would be impossible to separate the Egyptian from the modern Algerian. Tlie means taken above show a difference of only 2 millimetres in a dimension which differs by 14 millimetres in various races, and this is therefore insignificant. Breadth, maximum. — We here reach the commonest measurements, but those which have pioba])ly less intrinsic value owing to their complexity of the elements of growth in\'olved. We take both sexes together in order to compare with the Algerian tombs. The early prehistoric is 130, (male 131, female 128i, see curve 84); the later skulls give 132 (male 133-|, female 130|, curve 85). The Algerian tombs give 136 (male and female mixed, curve 86). The living Chawia males centre on 147^, with a break into two groups centering on 146 and 153 ; and the living Kabyles give 149^. From these probably 8 may be deducted for the flesh, and U for reduction to mixed sexes (see curves 84 and 85) thus the figures are : Early prehistoric . , , 130, Mixed (Naqada) ,,. 133. Late prehistoric ... 132. [8] Prof. AV. ^I. Flinders Petrie. — Migrations. Algerian tombs ... 136. Living Chawia ... 138, groups 136| and 143|. ,. Kabyle ... 140. (For the reduction of life measures to the skull, see Broca, Bull. Soc. Anthrop., 1868, p. 25.) Here it seems that the Egyptian skulls are 4 millimetres narrower than the Algerian. Length maximum. — The amounts, after allowing on the living 6 for flesh thickness and 3 for correcting to mixed sexes (see curves 80, 81) are as follows : — Early prehistoric ... 181|^. Later „ ... 181i. Algerian tombs ... 182. Living Chawia . . . 183. „ Kabyle ... 184. There is no difference of any significance between these figures, in view of the range of variation in each group. Or, if it be preferred to compound the length and breadth in an index, the cephalic index (reduced for the living to mixed sexes to agree with the Algerian tombs, and reduced for constant of living to skull forms, see also Pdpley, Races, p. 593), will be :— Early prehistoric ... 72, prob. error range 70-74^. Late „ ... 73 „ „ 71-75. Algerian tombs ... 75. Living Chawia ... 74. Kabyle ... 75. Thus the Algerian may be shortly stated as being at just about the probable error limit of the prehistoric Egyptian. And this difference of the means is due to about 4 milhmetres in the maximum breadth. We may, then, sum up the anthropological evidence for the Libyan source of the prehistoric Egyptian thus : — For; the 10 points of culture noted above, „ profile and colour, „ nasal height, and class of breadth, ,, nasi-alveolar height and bizygomatic breadth, „ the maximum length. Against ; „ burial customs, due to difference of soil, „ mixed origin of the historic Egyptian language, „ difference of 4 millimetres in the maximum breadth. The last datum is the only one that can be seriously placed against the large number of cultural and anthropometric points of agreement ; and the maximum breadth of the skull is perhaps the dimension most liable to va)"iatiou in a race by increased brain-growth There can be no reasonable doubt, after re\dewing all this e\"idence that the Prof. W. M. Fijxders Petrie. — Mirjmtions. [9] Libyan is the main stock of the Egyptian race in prehistoric times ; in accord with the obvious probability of the Egyptians being one with those cognate peoples which lie on either side of the country. The slight differences that we have noticed are far within the changes that may be expected from a difference in time of many thousand years, in space of 1,500 miles, and from the early mixture of one or more other races in Egypt itself. On referring to the curves 30 to 34, and their analysis there shown, it would appear that we must recognise two different groups in the early prehistoric age ; in the late prehistoric age these became fused ; and the Algerian agrees most nearly to the fused type of the facial height, and to a still later fusion of the facial width. Early. Late. Algerian. Nasal height ... 1 48 521 50 + 531 51 Nasi-alveolar ,.. ( 66 } 69 + 75 70 Bizygomatic widtJi 1 1221 129i } 126 130 It seems then that, as far as data so widely separated in time and place can be compared, there was a mixed race in North Africa and Egypt in the early prehistoric age ; and that this, fused together, has persisted in .Algeria with some slight improvement in general size, and especially the width of the skull from increase of brain. To get behind this mixed race is quite beyond our present data. That there was somewhat of the old palaeolithic Bushman stock is very probable ; and that there may have been another low type such as the Socratic Sinai Bedawy seems likely from its position. These results are, however, for the Abydos region ; and on going fifty miles further up the country to the Naqada region we find that the lower of the Ab^^dos types seems to predominate. Naqada. Abydos. Nasal height Nasal breadth Nasi-alveolar Bizygomatic width . . . 49 25 671 1251 A 48 27 66 122i B 521 24^ 73 1291 (Biauricular width was not measured on Naqada skulls.) [10] Pkof. AV. M. Flinders Petrie. — Mujrnlions. To settle how far either of these results may be representative is impossible until some other large series of prehistoric skulls may be obtained in different parts of the country. So far, it might well be that the Xaqada type had been mixed with a more European ty]je at Abydos, and also lower down in the Nile valley and along the African coast. The later prehistoric people were a fusion of the earlier elements, as we have noticed above, with a smaller addition — perhaps a third — of a higher type. That there was some distinct change in culture from about 38 to 44 sequence date,^ is evident from the considerable changes there. The older forms of pottery disappear at this time, and new foims come in. The plain red polished and the black topped yjottery cease to start new forms at 43 ; the fancy forms of pottery cease to arise after 40 ; whereas the decorated pottery with ships and animals practically begins at 40 ; the wavy-handled pottery also begins at 40, and the late rough pottery begins at 43. In short, the old style of ware, like the Kabyle, was arrested (the white cross-lined patterns had died out before), and the new styles had nothing in common with the Kabylian. The minority type of man with larger faces was apparently that of the people who brought about this change, as the few of this class clear of the majority range (curves 35 to 39) that can be dated are of sequence dates 42 to 65 or later, and probably were equally spread over the whole time. Such a continued separation points to their being a distinct class. This incoming people may be somewhat understood by the different character of objects which they brought in. The most important of these is the barrel- shaped vase of stone, unknown before s.d. 39, and iu full use by 42 sequence date. This form is shown in the I Dynasty as being offered in tribute by the people with pointed nose, and hair plaited in a pigtail, who also wear a long robe of skins. They came then from a rocky region where stone was used, and from a cold region where long robes were needed; yet not far from Egypt, as they were early subdued Ijy the dynastic race, and employed in the conquest of lower Egj^pt, shown on the slate carving with captives. Moreover, a few small vases of the decorated pottery are rarely found in earlier graves of 31-40 sequence date, probably imported, but showing that this other civilisation was in existence almost as early as any graves in Egypt. The only district which agrees with these indications would be the eastern desert hills. There are still many fertile valleys in this region, as at the convents of St. Antony and St. Paul, and the porphyry quarries ; it has been shown by Floyer that the eastern desert had much more vegetation before the introduction of the devastating camel ; and Sneferu is known to have nuide 122 tanks for cattle, probably in these desert valleys. A hardy people in this region might well obtain control of the Nile. They were accustomed to ships, as figures of great galleys are a common subject of their vase painting ; and model boats made of similar pottery and colouring are found almost at the beginning of the series of prehistoric graves, apparently imported ' The whole of the prehistoric graves are divided into a scale of 50 part.s, which ai-e mnubeied from 30 to 80, ending in the reign of Mena, founder of the I Dyna.sty. Ynov. W. M. Flixkeks Petkie. — Miyrntionf. [11] there, as no similar clay occurs so early in J^^gypt. They had considei able connection with other lands ; at 38-4(> sequence date the clay beads of earlier times are supplanted by lazuli, serpentine, haematite, and silver, which show a foreign trade. They substituted the pear-shaped mace-head for the older sharp-edged disc. They brought in tlie use of spoons, of amulets, and of forehead pendants. And the ostrich and the aloe were familiar to them, as shown by their paintings. The plaiting of the hair hito a pigtail, shown on the I Dynasty carvings, agrees with the cessation of the long-tootlied combs which were so commonly used before S.D. 40 to fasten up the hair. The position of these people, as a small body of hardy mountaineers, explains how a minority could rule the larger bulk of the Nile i'olk, and yet not be lost by admixture. Another racial indication is found in the painted grave of Hierakonpolis. There a black man is shown holding a black cord which ties up three red men kneeling; he is threatened by two red men advancing with sticks. This shows that at S.D. 63 there were conflicts between a red and a black race above Thebes. We now reach the age of the dynastic race which seems to have come across from the Eed Sea by the Koptos road. The immense difference in culture is at once evident. Never previously had there been any passable figures executed of men or animals, accompanying the fine mechanical work of the prehistoric times ; at once now spirited carvings appear, with that minute ethnographical distinction which the Egyptian continued to retain through all his art. In the prehistoric time only a signary of linear marks was in use, mainly by the earlier people, and diminishing as time went on ; the dynastic people bring in a pictorial system of hieroglyphs well developed. This outburst of artistic work has left a magnificent memorial in the series of slate palettes covered with reliefs relating to the conquest of Egypt, and the great mace-heads of king Narmer. From these we can trace three other races who were conquered in Egypt. The heads of these are published in Joar. Aiit/i. Inst., xxxi, PI. XIX. They are— 1. The plaited beard type, with curly hair and thick nose. These are pro1)al)ly from a hot climate, as they wear no clothing; but the face is not at all prognathous like the negro, nor is the nose short. They are most like figures at Ibriz in North Syria. They were conquered early in the unification of Egypt, and are therefore not prol)ably on the north border. The only mixture of the known races that might produce this type would be a mulatto mixture of the pointed-nose type with the negro, having the beard and nose length of one and the thickness and curly hair of the other. The pointed-nose type has been dealt with as probably the ruling race of the later prehistoric age. 2. The tilted-nose type; with a short nose, sharp pointed, and tilted up below. This is the type of a captive chief who was probably of the Fayum district ; his hair is wavy, and the growth of the beard like that of the Bedawy. He might be a mixture of Semite with the next type. [12] Prof. W. 'SI. Flinders Petrte. — Migrations. 3. The forward-lieai'il type, with sub-aquiline nose, lank hair aud a beard growing out straight in front. This is strikingly like the men on early Greek vases (see Nehesheh and Defenyieh, PI. XXX), and we might well assign this to the coast people of the Delta, mixed with northern emigrants, and' akin to the Fayum people behind them. These men served in the army of Xarmer, at his conquest of the Fayum. Unfortunately, all our early material for measurement comes from the Abydos-Xaqada region, and there we cannot hope to find any trace or influence of these three types, as they belong to the lower country. From the difference between Abydos and Xaqada material, only fifty miles apart, we see how much change there might be at some hundreds of miles distance. The dynastic race is marked by a face wholly different from all the other types. The forehead and nose are in almost a straight line, the head massive, the ear large and flat, the nose straiglit with rounded tip and slight slope beneath, the jaw long and square. {Jour. Anth. Inst., xxxi, PL XX). A head which is clearly of this type is here published for the first time, PI. XIX ; it is a life study of one of the earliest kings made as a model for sculptors to copy. From these dynastic ])eople first appearing at Abydos and El Kab, and later conquering the Delta, it seems that they entered Egypt at the middle, probably by the Eed Sea road to Koptos. The statues of the god ^liu, which I found there, are the rudest and earliest known, and bear artistic figiu^es of animals, with pteroceras shells and sword fish which are probably Eed Sea produce. The evidence of the skulls of the Abydos region (curves 40 to 44) would not indicate any noticeable numbers of a different race, unless indeed they were so much akin to the pointed-nose type as to appear unified with them. This is possible from the physiognomy, though they were not exactly the same people. The declining ci^alisation of the II Dynasty seems to have been broken up by an invasion fi-om the south ; the head of king Sa-nekht, the founder of the III Dynasty, being strikingly Sudanese in type (see Researches in Sinai, fig. 48). That some invasion occurred at this point is certain from comparison with other periods ; no declining civilisation ever revives, as this did at the pyramid age, with- out an infusion of new blood. The mixture, however, may not have reached below Upper Egypt, as there is no distinct new type appearing in the skull measures (curves 45 to 49) which are at this period from Middle Egypt, Deshasheh and Medum. They seem to show much the same two types, of the older stock and the dynastic race mixed. The portraits of the IV Dynasty show that the old Libyan stock was dominant in the ruling caste by that time. At the close of the Old Kingdom in the YI Dynasty we find the carved button seals coming in, which mark a large infusion of a ruder race in the VII- VIII Dynasties. On looking at the skull measures (curves 50-54) it is seen that the majority of the people at Dendereh in Upper Egypt were a fusion of the two older types ; while a minority of about half the amount were of a lower type, with short nose and narrow head, just the same as the lower type of prehistoric people. Journal of the Anthropological Institute, Vol. XXXV T, 190fi, Plate XIX. Pkof. W. M. Flindeks Petrie. — Migrations. [13] Low prehistoric. VI-XI Dynasty, low. Nasal height 48 49 „ width 24-4 2G Nasi-alveolar 66 65-6 Bizygomatic 1221 121i Biauricuhir 112 111 Such a near duplication of measures (without the least i^ersonal equation in imitating the earlier type when analysing them) is strong evidence that it was the same race again entering the country. It is likely, then, that they were the lower stratum of Libyans, who had become more civilised in three or four thousand years, and had developed a curious fashion in these button seals, which are more akin to early Cretan design than to anything else. At the same time a different stock was entering the land from the Syrian side. The sphinxes and heads which used formerly to be termed Hyksos are certainly older. Lately the similarity between these and the heads of the kings in the later half of the XII Dynasty has led to their being attributed to that age. But it will be clear on comparison that the Egyptian kings have the characteristics in a much softened form, without any of the savage strength of the unnamed heads. (Compare Fig. 108 with 141-2 in Students' History of Egyjit, I.) This new type had a large round head, nose slightly aquiline and broad, sharp projecting lips, and rather prognathous angle. The facial muscles are always strong and heavy, and the cheeks wide and flat. There is no trace of Mongolian tilt or smallness of the eyes. This powerful race, which most likely came from Syria, gave the new in- fusion which started the XII Dynasty civilisation ; and the kings of that age evidently inherited a good deal of the new type. The measurements of skulls of XII-XV Dynasties (curves 55 to 59) show a new stock coming in a of a liigh type, wliile the previous stocks were fused together. As these are so far south as Abydos, it is unlikely that this change would be due to the Hyksos migration ; it is probably the effect of the previous race that we have noted becoming mixed throughout the country ; if so, we must allow for a very large number of immigrants, as the new high element is almost as large as the old fused stock. Possibly it was a type that was domiiuint toward a recessive type of the old stock, and every descendant of the newcomers took up their character. Just the same cliaracter is seen on comparing the curves of female skulls. The XII-XV break into two nearly equal curves, of which the low one is the fusion of VI-XII groups, and the higher one is added new. But the new high group is so large that it cannot be refeiied to a Hyksos element so far south, [14] Pkof. W. M. Flinl)Ei;s TETiaE. — Miassed on from the Oder to the Seine (i, 329), but were repulsed; 354, they are in Thuringia (iii, 36 ; I, 223) ; 370, .S0,000 came down to the Ehine and retired again (iii, 37) ; 406, large numl)ers were exterminated at Florence (iii, 263, 267); and others join in raiding Gaul (iii, 268) ; 413, they crossed the Ehine, but were checked (S.D) ; 440, they raided Belgica, but were defeated and went down to Savoy 443 (iii, 450 ; II, 110); 450, see Map, II, 97; 470, they were south of Swabia (III, 21); 493, westward in Gaul, Map, III, 1 : 499, they occupied down to Marseille and Aries (iv, 110). But, 500, are otherwise said not to have held the lower hundred miles of the Ehone (III, 323). 10. SuEVE. — This was the name of a group of lesser tribes, which in a.d. 20 were in possession from the Ehine to the Elbe, and some of them — the Hermonduri and Longobardi — east of the Elbe (Strabo, VII, i, 3). They included the Semnones as chief tribe, Hermonduri, Longobardi, Eeudigni, Aviones, Angili, Varini, Kudoses, Suardones, and Nuithones (Strabo, Tacitus, Pliny) ; and the limits stated, over the Hercynian forest, and out to the frontier of the Getae, seem to include the Marcomanni, Quadi, Narisci, Marsigni, Buri and Lugii, and are accepted thus (in B, 241). The Oherusci were also confederate (B, 176). Of these tribes there may be identified in the sixth century in the Scops tale, Longbeardas, 11. 66, 162; Eondingas, 50 ; Engles, 15, 71, 89, 123 ; Waernas, 52, 119 ; and Sweord-weras, 126. The name Sueve descended to the Sweben or Swabians. 9 B.C. the Marcomans pushed from the Main into Bohemia and expelled the Celtic Boii (B, 131). 5 a.d. 40,000 Sueves and Sigambri settled between Meuse and Ehine (S.D.) ; 10 a.d. the Suevi occupied between Ehine and Elbe (Strabo, VII, i, 3). About 18 the Gotones from the Lower Vistula conquered the Marcomans (P>, 176). After 20, the Marcomnns still in Bohemia, and Quadi in Moravia (B, 241), on to 50, when many tied into Pannonia (B, 241); 50, the Lugii in Silesia were Sueves (B, 241); 100, Ptolemy places Suevi- Langobardi on Ehine, also Suevi east of these; and at 110, Suevi-Angrivarii (i.^.. Angel waras, on the Angel branch of Ems, 51° 50' N.) ; 211, on the Main (i, 257) ; 406, with others exterminated at Florence (iii, 263, 267) ; and in migration through Gaul to Spain (iii, 346) ; 409, settle in Galicia (iii, 346) ; 450, see Map, II, 97 ; 454, in north of Pannonia (II, 20) ; 470, in south-west of Germany, and along LTpper IJanube, mingled with Alamanns (III, 20); 493, Map, III, 1; 568, broken bands of Swabians join Longobards (0, 185) ; 580, end of Suevic kingdom (0, 189). 11. Vandal. — This people were reckoned by Pliny to include the Burgundiones [24] Pkof. AV. M. Blinders Petrie. — Migratiam. Varini, Carini, and Gotones. Zeuss counts them as a branch of Hermiones. Tacitus calls them Germans, but it is probable that there was considerable Slavonic mixture (S.D.). Procopius states that they were kin to the Goth and Gepid (II, 213) ; 60, they occupied between the Oder and the Vistula (Pliny, S.D.) and perhaps up to Holstein (i, 2-42) : 170, they joined the ]\Iarcomans (S.D.) ; 180 (?) they passed into Panuonia (S.D.) ; 181, settled in Dacia (II, 215) ; 250 {Vj settled between the Marisia and Crissia (i, 295). On the north of the Danube (Peutinger) ; 271, crossed Danube and returned to Dacia (11,216); 277, brought into Britain (II, 217); 331, in Moravia and north-west of Hungary and pushed over Danube ; 337, much reduced and fled into Pannouia (II, 218) : 405-6 raiil into Italy and exterminated at Florence (iii, 263, 267) ; 406, ravage Gaul (iii, 269) ; 409, settle in Galicia with Saevi (iii, 346) ; 409, the Silingi occupy Baetica, but in 418 were extinguished by Goths (II, 223) ; 429, Gaiseric takes 80,000 with Alans to Carthage (iii, 402-11 ; II, 244) ; 455, Genseric raids Eome (iv, 5) ; 480, the Moors regain the coast west of Caesarea (O, 28) ; 493, see Map, III, 1 ; 536, some go to Mauretania (IV, 37). 12. Alax. — AVe now leave Germans for the Sarmatians, Tartars, or Turanians (I, 739). B.C. 80 the Alans were in Alania and remained till the sixth or tenth century (S.D.). In they were among peoples north of Maeotis (S.D.) : 50, Seneca names them on the Danube (S.D.); 70, ravaged Media and Armenia (S.D.) ; 70, on Tanais and Maeotis, Josephus ; 100, Ptolemy puts the Alauni up far north of Crimea, apparently their original home ; 130, raiii Cappadocia (S.D.) ; 242, in Thrace (i, 190) ; 370, Roxalani, Maj), I, 185 ; 372, between A^'olga and Don, some to Caucasus, some to Baltic (iii, 91). After 375 Alans about Dniester (S.D.); 382, with Goths in Moesia, and mixed with them later (S.D.) ; 400, exterminated at Florence (iii, 263, 267); 406, raid Gaul (iii, 268-9). About 420 settle in Carthagena and Lusitania (S.D. ; iii, 346) ; 429, Gaiseric takes them, 80,000 with Vandals, to Carthage (iii, 402-11 ; II, 244): 450, settled in Valence (iii, 449 ; II, 111) ; 483, an edict addresses bishops of Vandals and Alans in Africa (S.D,). 13. Iazyg. — These were Sarmatians, akin to the Alans ; but they appear to be distinct from the Slavs (ii, 563 ; S.D.). In nature — and possibly in name — the Cossack is their modern representative. The first site of this people at the time of Strabo (VII, iii, 18) and Ovid, was north of the Crimea, marked 0. About 30 they reached the Theiss (ii, 216); and at 50 they are found holding the land between Danube and Theiss (B, 241); 335.. the Slavs over whom they ruled expelled them into Pannouia and among the Quadi (ii, 219-20), and 300,000 were accepted as colonists in Pannouia, Thrace, Macedonia, and Italy ; 357, they returned from the Quadi to their old land (ii, 263); 365, raiding Panuonia with the Quadi (1, 142) ; 950, they are known in Podlachia (S.D.). And after 1000 the Yazwingen occupy province of Lublin (Droysen Atlas). At present the people of the old district between Danube and Theiss are known as Yazygien, with towns Yasz Apathi and Yasz Bereny on the river Zagyva. 14. Slovane. — This was the original form of tlie name (Miklositch, Gram Slav. Sprache). In contact with Germany there were three tribes, the Venedi Pkok. W. M. Flinders VKTiUK—Mif/rations. [25 ] Slavi and Antae ; or tlie Wiiiidi were divided into Sclaveui and Antes (see Jordanes). The various tribes, where known, are marked on the map with V, S, and A respectively. In lUO the Venedi were to the east of Danzig according to Ptolemy, to which Tacitus well agrees ; and the Borusci (marked B) at some distance to the north-east. They were probably of the same stock, as they joined with the Wends later as Prussians. 300, the Antes were apparently in this region when met by the Longobardi in Anthaib (V, 94) ; 343, the Slaveni were on Upper Danube and Vistula, and the Antes between the Dniester and Dnieper (I, 77). The Venedi had pushed south, as by the third or fourth century there were Venadi Sarmati in North Dacia, and Venedi at the Danube mouth (Zeuss, 592) ; 360, the Venedi were in the Polish plains (iii, 58) ; 420, the Slavs (Wiltzes, Sorbes, Abotrites and Czechs) were on the Elbe aud Oder (0, 177j : 4:54, Slavonians, on break up of Attila's army, entered Carinthia and Carniola (II, 194) ; 470, Sarmatians got Singidunum, and lost it 472 ? (Ill, 24) ; 537 about, Antes from Dniester to Dnieper (I, 77; Jord.) ; 537, Antes in Moldavia and Wallachia (iv, 347) ; 549, Slavonians ravage Thrace and lUyria (IV, 564) ; 550, Slavonians attack Naissus and Thessalonika (IV, 567), and are turned into Dalmatia; 568, they enter Italy with the Longobards (0, 185) ; 582 to 7, they settle in Thrace, Macedonia and Greece (0, 153), and are driven back 601 ; 585, they liegin to move across Danube and settle in Moesia, which was almost desolate (O, 151, 152); 605, Slavs with Avars attacked Constantinople (0, 155) ; 610, Slavs on south-east of Alps and in South-East Bavaria (Zeuss, 616) ; 630, Venedi and Sorbi border on Thuringia (0, 179 ; iv, 296) ; 659, Slavs hold all Moesia, and inland Thrace and Macedon, to near Hadrianople and Thessalonika (0, 240); 770, Czechs and Moravians on Upper Elbe, Carentanians on Drave (0, 345) ; Slavinia east of Elbe, aud Abotrites up to Lubeck (0, 350) ; 800, Abotrites in Mecklenburg, Wiltzes in West Pomerania, Sorbes in Brandenburg (0, 360) ; Slavs on the Save and the Drave (0, 362) ; 804, Aljotrites extend to Eider (0, 366) ; 843, Slavs extend from Baltic to east of Istria (0, 410) ; 1000, Slavinia from Elbe to Oder; Polonia from Oder to Vistula (Droysen Atlas). Modern V, north of Denmark, Vend-syssel ; W. Wendelso ; SL, Slavonia, south of P)ayreuth (Kipley, Races, p. 244). At number 770, Mecklenburg, the Grand Duke of which is " Prince of the Wends." 15. Hun. — In beginning to deal with the Tartar races it should be noticed how all the Hun tribes have names compounded of Igur.* The Uighurs were eastern Turks, in the seventh century north-west of Mongolia (about 43°-46° N., 81°-95'^ E., Map, II, 1; and Skrine and Ross, Heart of Asm, p. 116); and the persistence of the name in nearly all the Hun tribes is a strong evidence of their origin. The names are in Jordanes, Priscus, Procopius, and Agathias. Kotr — igur. Ultz — inzur, Cutzi — agir. Uti — igur. Aniil — zur. Altzi — agir. Ult — izur. Alpil — zur. On — ogur. Ale — idzur. Sar — agur. Hun — ugur. Ac — atzir. Sat — agar. Bit — ugur. [2(1] Prof. W. M. Flinders Petrie. — Mifj/ratiom. Some of these may probably be duplicate forms of the same name ; l)ut all appear to lie compounds of Uighur, with softening of the g to j in some cases. Whether this is again connected with their old border river the Jaxartes (Yajar- tes) must be left open. For the large question of their identity with the Chinese Hiong-nu, see II, l-.'^)6. At 100 the Chuni were north of the Crimea, and it would be very difficult to dissociate them from the Huns ; the possibility, however, of a later addition to Ptolemy's geography might be a question. The first of the great migration drove out the Alans in 072. Thence they crossed the shallows of the sea of Azof (according to national tradition, I, 243) ; and in 374-6 burst upon the Gothic kingdom of Hermanric, and drove the Goths up the Danube to take refuge in the Ponian empire (1, 246-9) ; 378, they invaded Thrace (I, 264) ; 395. another branch raided Armenia, Cappadocia, Cilicia, and on to the Orontes (ir, 38); also the Upper Eujdirates (I, 654) ; 408, they cross the Danube, far into Bulgaria (II, 38) ; 425, 60,000 advanced to Italian border, and then settled in Pannonia (iii, 417). About 430 they ravage Media (iii, 423); 446, Moesia and south of Danube given to Huns (iii, 430) ; 451 Attila invaded Gaul, byway of the Danube and Main, as boats were built in the Hercynian forest (II, 108); they raided Belgic Gaul (II, 114), Metz (116), failed at Orleans (121); turned to Troyes, near which the great battle took place (124). Thence returned across the Pthine to Hungary, probably by a different route in order to get more spoil ; 452, moved from Hungary to Aquileia, and up the Po to Milan, and returned (II, 146- 169) ; 454, expulsion of the Huns from Hungary, after death of Attila, to Scythia (II, 193) ; 539, Huns raid to Constantinople and Corinth, and across into Asia (iv, 347-8, 537). U, Utigur Huns, K, Kotrigur Huns dwell divided by the Don (iv, 537) ; 550, Kotrigurs were brought by Gepids against Longobards ; 551, Uturgurs attack Kotrigurs, enter Crimea and across strait of Kertch (iv, 538); 559, Kotrigurs cross Moesia and up to Constantinople (IV, 52o-532.) 16. Avar.— These were kin to the Huns, living under Khans (S.D.); 558, migrated to the Alans at foot of Caucasus (iv, 353); 559, on Don and Dnieper (v, 516); ."60, defeated the Antae ; 562, fought on the Elbe (v, 516). Probably centred in Galicia at 565 (v, 517); 567, spread all over Gepid lands, occupying all west of Theiss; 570, held the Danube from Vienna to mouth, and ravaged Balkans (0, 146) ; 581, conquered Slavs of Dacia and Lower Danube (v, 517) ; 605, with the Slav raid to Constantinople (0, 155); 619, raids up to Constantinople (v, 75) ; 770, on the Drave and Pannonia (0, 345) ; 788 to Friuli (v, 74; 0, 361) ; 800, decaying on the Danube (0, 361) ; 896, decaying and mixed with Slavonic Moravians (0, 471). 17. BuLGAR. — These were also an Ugrian tribe (0, 248), and part of the Huns (Zeuss). About 450 between Danube and Dnieper, probably a stranded fragment of Attila's army. Note that in 765 Irnek, son of Attila, is claimed as a Bulgar prince* (vi, 546) and his date agrees to the period. 570 (?) they formed part of the Avar empire (vi, 130); 635, they were still north of Danube (vi, 545); 640 (?), they cross into Pannonia and on to Adriatic ; and they reigned south of the Danube Prof. W. M. Flinders Petrie. — Mir/rations. [27] from 640 to 1017 (vi, 136); 679, settle in Moesia (vi, 545); subdued the Slavs aud occupied up to the Balkans, adopted Slavonic (0, 248); 680, five centres of Bulgars and kin, (1) Don to Dnieper, (2) Kotrigurs beyond Don. (3) Danube kingdom including Utigurs, (4) Pannonia, (5) settlements in Italy (vi, 545) ; 690 (?), 30,000 taken captive, and sent to Asia (0, 249), where they joined the Saracens (0, 250). [This great weakening of the Bulgars may have been the cause of Slavonic prevailing after.] 750 (?), by the sea of Azof, west of Don (vi, 545); 550 to 750, extension of Slavonic settlement in Greece owing to war and plague (in 749) clearing the land (vi, 543); 950 (?), black Bulgars far north between Volga and Kama; Bury supposes migrants from Azof, but may have been left on the same incoming track as that of the Huns. 18. Ugur. — The Ugri or Ungri were of Finnish stock in the main, with Turkic added (vi, 552). From tlie name it seems impossible to separate them from the -ugur tribes of the Huns,* and when they first appear, 837, they are called Oungroi, Ounnoi, and Tourkoi. About 834-7 they were in Lebedia, near Chazaria (vi, 552) ; Bury supposes it to be between the Don and Dnieper, but this should be compared with Livadia* in south of Crimea. 837-9, they cross the Danube in a raid, their home being perhaps nearer than before (vi, 552) ; 840-90, they were in Atelkuzu between Dnieper and Danube, and 896 entered Hungary (vi, 552). By 899 they raid at Verona (0, 465); 907, they overrun Danube up to Linz (0, 474); 913, up to Coblenz (0, 476); 917, to Basel (0, 476); 924, Pavia looted and burned, and annual raids extended to Calabria (vi, 143); 924, raid up to Constantinople (vi, 144); 955, raid into Flanders, but finally checked by Otho (vi, 145-6). 19. Turk. — This race approached by the south of the Caspian, and not by the north, like all the preceding Asiatics. 841, the Turkish guards were brought in to Baglidad, and about 850 there were 50,000 established in camp at Samara (vi, 47) ; 1050, raid from Taurus to Erzerum (vi, 235); 1063, cross tlie Euphrates and take Caesarea; 1065, take Armenia and Georgia (vi, 236). Driven back to Euphrates in 1070; 1074, they were camped in Phrygia, and by 1084 occupied Asia Minor (vi, 248) ; 1076, take Jerusalem (vi, 206). After a pause of over two centuries the Othman Turks renew the conquests. Seated in Phrygia about 1300 (vii, 23), by 1312 they take Ionia and Lydia (vii, 26) ; 1326, they take Prusa (vii, 24) ; 1360, Thrace was subdued ; 1361 Hadrianople taken ; 1362, Philippopolis (vii, 31); 1373, Macedojiia occupied; 1375, Bulgaria became a vassal; 1385, Sophia was taken (vii, 31); 1453, Constantinople taken; 1460, the Morea taken (vii, 204). 20. Arab. — In the Notitia (402) there are Saracen cavalry of the tribe of Thamud stationed between Cairo and Belbays. The active movement, precipitated by Islam, began with the taking of Damascus in 633 (v, 419) ; 637, Jerusalem was taken (v, 434) ; 638, Aleppo and Antioch taken (v, 437) ; 639, Egypt occupied (v, 445); 641, Alexandria taken (v, 451); 647, Safetula was taken (v, 462) ; 654, Khodes taken (0, 239) ; 659, most of Armenia taken (0, 240) ; 668, siege of Constantinople (vi, 2) ; 683, Larnbesa taken (v, 464) ; 698, conquest of Carthage [28] Prof. W. M. Flinders Petkie. — Migrations. (v, 469); 710, Spain entered (v, 474); 711, Sardinia taken and Tyana (0, 254); 712, Pontus taken (O, 254) ; 713, all Spain (except nnrtli-west) conquered (v, 480), and colonised thickly from Arabia, Syria, and Africa (0, 506) ; 713, Pi.sidia and Asia Minor in general occupied (0, 254); 716, siege of Constantinople, and death of 80,000 (vi, 5 ; 0, 303) ; 731, Septimania taken (vi, 14) ; 732, defeat at Poictiers (vi, 17) ; and slaughter of 375,000 (questionable, 0, 295) ; 739, pushed back to Narbonne (0, 296); 751, driven out of the plain of Leon (0, 507); 790, they lose Sardinia (0, 364) ; 797, evacuate Barcelona and all north Spain (0, 365) ; 823, they conquer (Jrete (vi, 37) ; 827, they enter Sicily (vi, 38 ; 0, 448); 831, Palermo taken (vi, 39) ; 843 to 851, south Italy occupied (0, 450); 846, raid on Kome (vi, 40); 853-71, the Sultanate from Bari to Eeggio (0, 452); 878, Syracuse taken (vi, 39) ; 882, settle at Garigliano and raid Samnium (0, 461); 890, raids extend up to Lausanne (0, 464) ; 902, Taormina taken (vi, 39). Having now placed the successive movements of each people together as a consecutive view of their migrations, another view may be taken of the subject, referring to migration as a system, apart from the importance of the peoples separately. The maps 21 to 28, here given, show the direction, distance, and date of the movements, in order to compare them. It is obvious that a displace- ment of one people is likely to cause another movement, either in front or in rear, or both. Hence the general direction of migrations may be expected to be similar in any one age. The use of such maps is va.rious. If a movement be doubtful in its place or its time, the resemlilance of other movements will help to prove or refute it. If the date of a migration is uncertain, the date of others affecting the same region may settle it. If it be doul:)ted whether a people can have come out of a region, the example of other movements may establish its proljability. The division here is roughly into centuries, but the intervals of tranquility have rather been taken as dividing the groujjs of movements. 21. B.C. 29-90 A.D. — This shows that two movements were in progress : one a flow down from the west side of Denmark, pushing even eastward from this, and carrying forward down to the Danube; another move was part of the ceaseless flow of eastern peoples into the Danube valley and Hungary. 22. 100-211 A.D. Here the northern flow is very strong. The great line down Denmark continued, but there was also a line directly across the Baltic. This led to serious pressure on the Poman empire, which was checked by Aurelius, and the great Gothic trek down to the Euxine. 23. 238-287 a.d. The eastern pressure now began, driving the Germanic peoples to the west and up the Danubian highway. A few dates here given are not entered on the racial maps. 238 the Carpi from the back of Hungary pushed down to the Dobrudsha (I, 48); 277, about, the Burgunds pushed the Alernans from the Main to the Rhine (i, 463), the date is probably given by their passing on to the Seine, whence they were repulsed (i, 329) ; 286, the Alemans and Burgunds both invade Gaul, probably by different routes (i, 463) ; 268 or later, the Goths push into the Crimea (i, 261) ; 277, the Franks were driven back into the morasses (i, 329). Prof. W. M. Flinders Petrie. — Migrations. [29J 24. 331-395 a.d. The great movement of the reign of Gallienus having been checked, there was quiet for a couple of generations so far as the Roman Empire was concerned. But the break-up of (Jonstantine's rule brought on a fresh pressure from the east. The easternmost of the Germanic tribes, the Burgunds, had moved west from Prussia to the Ehine, and the Alans pushed into their place. This apparently marks the loss by the Germans of the country east of Denmark : the pressure was all eastern, flowing into Gennany, along the Danube, and even past the Caucasus. 25. 401-493 A.D. P>y this century the central Asian Huns had pushed through to Europe ; and the pressure was so severe that whole peoples were projected through the Koman Empire into Spain and Gaul. There seems to have been some northern pressure also, by the Gepid in 451 and Sueve in 454 flowing south-east, as well as the south and south-west movements on this side of Denmark. 26. 512-589 a.d. The newly-settled peoples west of the Danube had crystallised a firm resistance to further pressure ; and thus the Avar flow mainly broke into the Greek peninsula. 27. 605-698 a.d. Another Asiatic wave — that of the Bulgars — pressed the Danubian region, which overflowed into Italy. The great southern movement of the Saracen (or Sharuqin " easterners "\ now broke up into Syria and along north Africa. 28. 710-804 A.D. — Tlie great movements from Asia now cease; the Arab pushed into Asia Minor and up Spain and France. The Slav made a final move westward up to the Elbe mouth, and also south into full occupation of Greece. The movements after this are outside of the scope which I proposed ; but they can be seen in their principal importance in the maps of the Ugur, 18, and the Turk, 19. Across the latter one would have to be brought in the considerable counter movement of the Crusades, a continued shift of tlie Germanic peoples south and east, but not on a national scale. After this review of the historical anthropology, we may turn to see if it has a serious bearing on the interpretation of anthropometry. I cannot profess here to discuss all the results which require to be treated ; but in one respect the historical evidence has been so much slighted that it requires a more decisive statement. In The Races of Europe Eipley argues about tlie Lombard invasion : " Eighty-thousand immigrants in the most thickly-settled area in ancient Europe surely would not have diluted the population very greatly" (p. 254). Now the whole historic evidence goes to show that so far from Lombardy having been " the most-thickly settled area," it was practically empty when the Lombards entered it. Tlie decay of the Italian population had long been the disease of the empire ; in 380 Ambrose deplored the ruin of the lower Po valley (iv, 55) ; in 480 Gelasius wrote that in Aemilia, Tuscany, and adjacent provinces, the human species was almost extirpated (iv, 56). Tliis void was filled by about lialf a million Goths in 489. In 536 there were 200,000 in the Gothic army. Ten years of incessant [30] Prof. W. M. Flinders Petrie. — Migrations. war up and down the whole peninsula, between the Goths and the extraneous array of Justinian (lY, 497), must have wrecked all the remaininp; chances of settled population ; besides a plague in 542 which is reckoned to have destroyed a third of the population (( ), 94, 181). The Goths were reduced from 200,000 to 20,000, and after that had been further diminished, the Inst remnant retired from Italy with all their families aiul possessions (IV, 656). Then 75,000 savnge Franks and Alamans raided down the whole length of Italy, destroying everytliing that mi.u;ht reniain, aa far as Sicily ; and then they were all exterminated by pestilence and a final slaughter of 40,000 by Xarses (0, 106). After these successive rinsings out of the land, especially cf the northern region, it is no wonder that the Lombards found on coming in that " the land seemed to have sunk back into primeval silence and solitude " — as their national historian Jordanes states. And the conclusion of modern historians is that " all the northern parts of the peninsula were desolate and well-nigh unin- habited " (0, 181) ; " alike in the northern plain, in Piceuum and Aemilia, and in the neighbourhood of Pome, the whole population had disappeared" (0, 106). Theie was a small imperial garrison left in the fortresses, but only three cities in the whole of the Po valley attonpted even a passive resistance ; and " the whole Lombard nation — men, women, and children, with their cattle and slaves — descended into the Venetian plains and spread themselves over the deserted lands " (0, 184). If ever we could verify the wiping out of an earlier population by decay, and then Ijy two successiA'e rinsings out by different peoples who were each exterminated, it is in this case. vSo far from being " the most-thickly settled area in ancient Europe," it was probably the thinnest. Now let us remember that these Lombards had left Scandinavia, which has the longest-headed population of Europe, and yet at present the Lombards have almost the broadest heads in Europe. "VVe are reduced to enquire (1) whether Scandinavia has entirely changed its type, or (2) whether the Lombards have entirely changed, or (3) if the Lombards who entered Italy in 568 had any resemblance to their forefathers of three centuries earlier in the Baltic plains ? To my own sense of history it seems certain that twelve hundred years have sufficed to change entirely the cephalic index of a people so as to accord with their environment. I have dwelt on this at length because it is a very clear case of historical fact ; but it does not stand alone in its teaching. Pipley discusses the Jewish cephalic type {Races, p. 397) : " In long-headed Africa they were dolichocephaKc. Ln brachy cephalic Piedmont, though supposedly of Sephardim descent, they were quite like the Italians of Turin. And all over Slavic Europe no distinction in head-form between Jew and Christian existed. In the Caucasus also they approximate closely the cranial characteristics of their neighbours." And then adopting the big petitio principii that cephalic index is constant in race, he concludes that therefore the Jews have everywhere become swamped by an included majority of Christians, and this in the face of tlie Ghetto and all its obloquy. Surely history and common observation lead us to the equally legitimate conclusion that the counlry and not the race determines the cranium. Prof. W. M. Flindeks Petrie. — Migrations. [31] To take another case, not so historically complete. All our evidence shows that Europe was lastly divided, by a line northward from the Adriatic to the isles of Denmark, into a Slavic east and a Germanic west. Yet the cephalic index divides Europe exactly across this, into a long-headed north, and a short-headed south. I fail to see a single proved case of racial permanence of cephalic index, while environment has changed ; but each case tliat we can test shows a local permanence of cephalic index while tlie race has changed. This is a radical question to be settled as to the meaning of anthropometry, and a question which can only be settled by historical evidence. A prion, the cephalic idex is just as likely to depend on place as on parentage, after a sufftcient time has elapsed for the conditions to take effect. To turn to another supposed test of race, Nigrescence, or Brunetness, or more simply — Colour. This subject has been studied in adniiralile detail, as everyone will know from the Huxley lecture of last year. But looking at the maps of colour in Europe, we see a distribution which shows scarcely a trace of difference at the real dividing line from the Adriatic to the Baltic, separating Slav from German ; whereas looking north and south we find large differences in either the Slavic or the German region. As a whole, colour approximates far more to iso-thermals than to iso-racials, a result to which Dr. Beddoe has already called attention. It seems, then, that two of the main physical tests, cephalic index and colour, must be only regarded as decisive during periods of less than a thousand years. A difference existing under like conditions, such as the diversity within a small area, is good proof of a mixture within a thousand years. But a difference or a similarity in different regions proves nothing for longer periods. To review in brief our anthropological tools. (1) Bony structure is certainly subject to the influence of conditions within a few centuries ; but a priori, it seems better to study single bones, where there is only one factor of change, than compound forms, such as the whole skull where many factors are mixed with variable co-efticients. (2) Colour is not more stable than structure ; perhaps the colour of well-protected parts, such as the armpit, would be less liable to vary by conditions. (3) Physiognomy and general expression is very vague, and ini])ossible to compress into fonnuhe with any definiteness, as two similar descriptions might mean entirely different appearances. Yet the general type of features is very marked ; and the expression can be estimated by so far more delicate and complex mental analysis than any measurement can give, that it deserves much more systematic study. Tests of discrimination of mixed photographs should be systematically carried out to ascertain the value of facial details and of mental estimates. In general, the possible scope of anthropometry on the least utilitarian structural variation, needs to be carefully investigated. Such a detail as the brain venation shown inside the skull promises, a priori, to give variation of descent irrespective of conditions. , (4) Language, which has been too much trusted or distrusted. We require a study of the effect of known mixtures and amalgamations [32] Prof. W. ]M. Flinders Pethie. — Migrations. on the structure, on the common words, and on the general vocabulary. In particular, place-names need a more exact valuation racially, as they seem to be by far the most valuable department of language. (5) Culture, which may be borrowed, but which has enormous value owing to the variety in it giving absolute proofs of a connection. In this class of evidence. Laws, especially of descent, are of the first value, as a people cling to them most closely, and they involve property. The Arts are so linked with the mental structure that they vary but little in character with changed conditions. The utilitarian types of houses, tombs, and clothing may vary greatly with conditions, but yet are of much weight when alike in small detail. (6) History, which we have here endeavoured to bring more into the scope of evidence, is so absolute that it is of greater weight than all other proof, where we are aljle to distinguish between political and racial domination. I greatly hope that the preliminary arrangement of the facts which I have here attempted, may lead students to more completely put in order every detail of historical record that has been preserved. Lastly, let us look at the meaning of migrations. We deal lightly and coldly with the abstract facts, but they represent the most terrible tragedies of all humanity — the wreck of the whole system of civilisation, protracted starvation, wholesale massacre. Had the Goths been left alone in their humane occupation of Italy, there might have been a set-back of two or three centuries ; but, by expelling them, civilisation was thrown fifteen centuries back. This fearful waste, not only of life, but of all the best gains and endeavours of man, if it cannot be avoided, shows hideous incompetence ; if it can be avoided, it is then the most gigantic of crimes. Can it be avoided ? That is the question before all others to the statesman who looks beyond the present hour. If we have to sum up the problem in one word, that is, "Weeding. The one means by which all natural progress has been gained is weeding : the only means by which impioved races have come forward in the world is by weeding. The hardest weeded race, which has endured most, has always overcome the less weeded race. The fatal curse of Eome was the state-maintenance of a people among whom weeding was thus at an end. The maximum of opportunity to the most able, the full penalty of incompetence when deser^•ed, is the only rule for a state which intends to avoid the far more terrible fate of a catastrophe when it touches a more competent people. The most recent panaceas of political ignorance, equality of wages and the right to maintenance, are the surest high road to racial extinction. The higher the walls of artificial restrictions, — the exclusion of more industrious races, the limitation of free labour, the penalising of the capable in order artificially to maintain the incapable, — the more certain and more sweeping will be the niigration of a stronger and better race into the misused land. The one great lesson of all this world-agony of migrations is the necessity of weeding ; and the statesman's duty is to see that this is done with the least disturbance, the least pain, and the most whole-hearted effect. FRAN Journal of tlie Anthropological Institute, Vol, XXXVI., 1906, Plate XX *^fe. ' ^j^a^. ■i\. "^ / :-^4H#'/"'''>^ to «i 1 1( ! ,>^ SAXON / '.>'- ^/ MIGRATIONS. Bale b Daniel^eoo, Ltd., Fhoto-btk Journal of tJie Anthropological Institute, Vol. XXXVI., 1906, Plate XXI. L0N60BARD «X?^f>,^ '\ 640 r'> MIGRATIONS. Bale & DEnicIiBan, Ltd., PhotO'litta Journal of the Anthropological Institute, Vol. XXXVL, 1906, Plate XXII. '^:r'^^ {hy^j^99^ r? i f >< MIGRATIONS. Rale % Danielasor., Ltd., PhotuJith Journal of the Anthropological Institute, Vol. XXXVJ.,1906, Plate XXIII. 13 lAZYP 14 SLOVANE< e^ .,v ^■ssir Lrz .« .*^>t^i 550V-, i ^ -■^ '1 //^5j '-»-»»* /C' f' MIGRATIONS. Bsle ft Daitirlison, Ltd., Photo-bti Journal of the Anthropological Institute, Vol. XXXVI., 1906, Plate XXIV. MIGEATIONS. Bale & DanieUson, Ltd., Photo-bth Journal of the Anthropological Institiite, Vol. XXXVI., 1906, Plate XXV. .C.29 90A0 ^~.t-/,/>>A -*»JJ,\_ L MIGKATIONS. Bsl«& DamtUsoo. ltd., Ph»to-b-.: Journal of the Anthropological Institute, Vol. XXXVL, 1906, Plate XXVI, j>f^!m^'^'.^ 'Ji 'i.^'-i^-^. MIGRATIONS. fwXe k lUoK-lKon, ItJ., Photn-iltr Pkof. W. M. Flindeks Petrie. — Migrations. [33] APPENDIX. The Interpretation of Curves. [Plates XXVII- XXIX.] A. The BaJaiiciiig of Custialtg. It is desirable in all cases to gain an insight into the nature of distribution by means of curves and graphic treatment rather than by inspecting numbers alone, even when the amount of material may be enough for numerical mathematical analysis subsequently. But the principles of the interpretation of curves seems scarcely to have been noticed by anthropologists ; this is much as if inscriptions were freely translated without any study of grammar. Moreover, the need of principles in graphic interpretation is all the greater because the amount of material generally to be oljtained, of homogeneous character, is very seldom enough to fairly render results from a mathematician's point of view. AVe are often loftily warned that no certain results can be obtained from any but a large amount of examples — a hundred at least — while we remain in darkness as to how mucli we can possibly extract from the few dozen of instances which is all that can be secured in most cases. The practical way to learn how to deal with small groups of material is to make several small groups from the material of one large one by casual selection. For an instance of this we may take a large group of over 150 examples of nasal breadth of Egyptian male skulls of the VI to the XII Dynasties from iJendereh. Any other large group would do as well, only this one is of importance in itself, as it lies at the basis of the results in a large work on the subject. The total of the whole group is shown in the curve Xo. 1, formed by 167 examples. The question arises, Is the departure of this curve from the normal curve of distribution to be taken as evidence of heterogeneous origin ? To study this, let us suppose that we only had a chance third of the examples, such as the 3rd, 6tli, 9th, 12th ... in the whole list; similarly we may take the 1st, 4th, 7th, and 10th, . . . and the 2nd, 5th, 8th, and 11th, ... as giving two other chance series. Draw the curves from these three groups and we have the curves in Xo. 3. Here each curve is of more than fifty examples, a larger amount than is generally available in anthropometry. lUit it will Ije seen that not a single departure from the normal in one curve is common to the other two. Every obvious variation is therefore due to casualty. And as the variations of each of these curves from the normal is larger than that of the total curve, and they do not agree in the abnormalities of the total, they sliow that there is little reason to attribute the abnormalities of the total to any cause beyond casualty. Now let us follow this line further. These abnormalities of the three curves being proved to be casual by their disagreement, this is equivalent to saying that there are too few instances in each division of the scale to uive a fair average for [34] 1'i;ijf. AV. 'SI. Flinders Petrie. — Migmtions. each division, or the scale is too much subdivided. To take an extreme case, it is obvious that if we had even a thousand examples, but insisted on too many subdivisions — say live liuudred — we should not get an even series of totals in each division forming a smooth curve. The number of divisions of the scale must bear some proportion to the total of examples to give a result free from casualty. Here ten divisions for fifty examples, or five per division, show abnormalities ; and even fifteen prr division in the total curve Xo. 1 is probably too few to smooth out all abnormality. If, then, we have but few examples we must adopt but few divisions : and in the curves in Xo. 4 the same fifty examples are divided only into a scale of five divisions, or ten ]?(v division. The results are much more harmonious, and sliow little departure from the normal. On looking at it from another side, let us see liow reducing the numljer per division acts. In Xo. 3 there are five examples per division, in X'^o. 5 are three examples, in Xo. 7 two examples, in Xo. 9 only one example per division. The increase of abnormality and of differences between the three curves is very obvious as we descend. Similarly in Xo. 2 there are twenty examples ^x/* division, in X'^o. 4 there are ten examples per division, in Xo. 6 are six examples, in Xo. 8 are four examples, and in Xo. 10 are two examples per division. From all these we may well say that nothing less than an average of ten e^'Cimples ^ev division of the scale employed loill gice results fairly free of casual abnormality. If, then, we have only twenty or thirty examples we cannot adopt more than three divisions to the scale, fewer that can give any curve worth having. For so few instances it is impossible to prove abnormality by a curve. And to avoid Iteing misled by casualty we must ne\er adopt more divisions in a scale than one-tenth of the number of examples. The lower mode of graphica,l inspection by spots along a line may be useful, and may in extreme instances indicate heterogeneous origin. We may note in passing that the diminution of the number of instances has very little effect on the mean, and on the probable error of one example, either as graphically estimated or as calculated rigidly. The results from these curves were all estimated by eye separately, and they come out : — Estimated. Calculated. 50 examples Xo. 3 25-8 2o:) Xo. 4 25-86 on 164 ±1-5 ±l-o ±1-17 oO „ Xo. 5 26-0 25-5 Xo. 6 25-7 on 90 ±1-2 ±1-5 + 1-1 20 Xo. 7 2.V7 25-0 No.- 8 2. ".-6 on GO ±1-4 ±1-5 ±1-2 10 „ Xo. 9 2o-7 25-7 Xo. 10 25-8 on 30 ±1-0 ±1-2 ± -9 Prof. W. M. Flinders Petrie. — Migrations. [35] The general conclusions then are, that ten examples give as good a mean as larger numbers ; but that even fifty examples would not suffice to prove a minor abnormality or heterogeneity. To these conclusions a ridor must be added that we are only dealing with sucli material as anthropometric measurements, of which the variations of one source will overlap more or less the variations of another. In the abstract view, the number of examples needed to prove heterogeneity is a function of the probable errors of tlie two sources and their distance apart. Where the distance is many times larger than the probable error, less than a dozen examples may prove difference of source ; on the other hand, many thousands of examples would be needed to prove difference of source where the distance was only a small fraction of the probable error. There is a curious prejudice sometimes shown against " smoothing curves," by taking the averages of adjacent columns : it seems to be supposed that this is a " cooking " process, and that it opens the way to arbitrary alteration of results. It is quite likely tliat tlie adoption of wider spaces of the scale in order to avoid casual variations, as above described, may be resented as being equivalent to taking the averages of adjacent columns; a word is therefore needed about this fallacy. Such objection to averaging or adding adjacent columns together can only exist with a l)lind fetishism of the millimetre. Were the measurements all taken in hundi-edths of an inch, the results in millimetres would l^e e(|ual to adding or averaging four columns together of the hundredths scale ; or if the measurements were all in tenths of inches, they would equal adding or averaging two or three columns together of the millimetre scale. The only rational rule for determining what closeness of scale shall be adopted is to see tliat there shall be an average of at least five to ten examples in each division of tlie whole range of the curve, as we have already seen that this is barely enough to avoid gross casualty. If there be enough examples it is better to adopt a scale which shall give an average of ten to fifteen examples in each division. If there are so few examples that there are but three divisions possible, it only proves that there are not enough to give a curve of any value. Another point to notice is that if we are adding, say, three columns together of a very close scale (as if we were using measurements in tenths of inches instead of millimetres) there is no virtue in any particular mode of grouping ; ])ut the groups divided, as columns, 1, 2, 3, [ 4, 5, 6, | 7, 8, 9..., or 2, 3, 4, | 5, 6, 7, | 8, 9, 10..., or 3, 4, 5, I G, 7, 8, | 9, 10, 11..., are all equally valid. The outcome of tins is that if there are not about ten or more examples in each division of the scale or unit of measuremen.t employed on an average of the whole curve, it is better to average the columns in twos or threes, or even larger groups. The error caused by assuming that tlie probal)ility curve is straiglit along- short lengths is less than the error produced by casualty of distribution. And this averaging of adjacent columns should be repeated at every single unit of measurement along tlie scale ; the number of columns to be added together (or averaged) being c 2 [36] Phof. W. M. Flinders Petrie. — Migrations. fixed by the need of having an average of about ten examples in each compound eohnnn as used for plotting. B. The effect of Heterogeneitii. The actual effect of the mixture of two sources upon the combined mass of observations is so seldom seen, that it is well to review it and to draw some conclusions. The curves 11 to 29 are each compounded from two normal prol^ability curves, tlie separate forms of which are incliuled in the area. The three columns (of curves 11 to 25) show varying degrees of distance between the component curves ; the five rows show varying proportions of height between the component curves. A third variable is that of the breadths of the curves, and some examples are given in curves 26 to 29 : but that is far less important in this subject, as variation of any human dimension is not very different in different races, so the widths of the curves will be nearly the same. The relative amounts of the sources is the most varial >le element, and five grades of it are given here. The distance of the sources may also vary a good deal, but when further apart than here shown, in curves 11 to 15, the component curves are easily parted on inspection. The right-hand curve is the variable, the left-hand curve is uniform throughout. Upon both curves the vertex is marked liy a short line, and upon the left-hand curve another line shows the position of its half area, or of probable error. It need hardly lie stated that the normal curve of distribution, or probability curve, may be magnified indefinitely in height (number of examples) or width (range of variety), but the character of the curve is always the same, and the variations which it may undergo should be clearly in the memory. Oljserve that the middle third of the height is almost straight, and the point of greatest flexure is aliout one-eighth from the base. When the sources are al)Out four times the probable error apart they are easily located in the compound. Curves 11 or 12 woidd be analysed at once by inspection ; and 13 to 15 show how we are to interpret a curve with a lop-sided tail. When the sources have their probable errors just clear of each other, the result is almost indistinguishable from a single curve if they are equal in amount, as in curve 16. This shows that any lack of height, or extra width in a curve gives ground for expecting two equal sources involved. In the lower curves, 17, 18, 19, the slight humpiness in a curve is seen to imply a minor source of about half the amount ; and in curve 20 the slightly lengthened tail is produced l)y a source only a fifth or less of the main source. When the two sources approach to near the probable error of each, as in curves 21 to 25, it becomes impossible to distinguish them by eye from a single curve. A main lesson to be learned from these curves is that unless we are dealing with sources extremely different in their amounts of probable error — say as 1 to 10 — (which do not occur in anthropometry), we cannot attribute any sliarp peaks in a curve to heterogeneous sources. The cause of sharp peaks we have already seen PitoF. W. M. Flixdeks Peffje. — Migmtiom. \})~] in curves 3 to 10 to be entirely due to an insufticient number of examples in each compartment of the scale employed. In curves 2G, 27, aud 28 are shown the compounds of two curves equal in height, one half the width of the other, and at varying distances. And in curve 29 is a near approach to a flat-topped curve produced by two equal curves. C. The analysis of compound curves. Having now simplified the material by showing that a notched curve must be cleared of casualty, by taking wider units of scale, and having looked at the instruments of research in the various types of compound curves, we may try how fnr it is practicable to analyse the compound curves found in actual observations. For selecting the material most likely to give an intelligible result, I proposed in 1895 (Brit. Assoc), that single elements of growth were more promising than compounds of several elements, which might vary in opposite directions. Thus the dimensions of the whole skull are composed of so many separate elements of growtli, that it was very likely that the final result might hide some variations. Similarly the effect of compounding two amounts in an " index " may entirely mask changes in the difierent factors, especially where each factor may consist of two rather different groups. For analysis it seems best to try what can be done on the simplest elements of growth ; if we can interpret those it will be time enough to then interpret an " index " by means of them. Of course, the usual view is that proportion alone is of consequence, and larger or smaller growth is immaterial. But what first threw doubt on that was an attempt to carry it out completely, by adopting the cube root of the capacity as a modulus for all dimensions of a skull. The results were that the grouping visilile in the dimensions or indices entirely disappeared, and mere confusion resulted. Among the dimensions which are usually published, the simplest are the nasi- alveolar length (or height of the face bone), the nasal height, the nasal width, the l)izygomatic width, and the biauricular width. The first three of these each depend on a single boue or symmetric pair, and the latter two are almost as simple. On trying an extensive series of possible correlations of dimensions, it was in these facial measurements that Mr. Maclver found the best evidences of correlation, and the most intelligible results. I have used only those measurements which he has published in the tables attached to The Ancient Races of the Tlieba'icl, as that comprises most of the material whicli I had collected, and is available for anyone to examine the results here given. The curves of distriljution of the examples of these five dimensions in nine successive periods are shown in curves 30 to 74. In these curves the outer one is the compound curve of actual examples ; the dotted curves within that are the components of it, as deduced from the study of these curves. On the left hand of each curve is a number stating the average number of examples in a unit of scale. Where the number is low a jagged curve must be expected. And the use of this number is when judging of the smoothness and freedom from casualty in a cur\c. c 3 [38] Prof. W. ]\I. Flinders Petrie. — Migrations. For instance, the nasal height curves, Nos. 30, 60, 65, and 70, are all mucli smoother tlian the corresponding nasi-alveolar height cm-ves,Nos. 32, 62, 67, and 72, although the numher of examples per unit in any of these pairs is closely alike. This indicates that nasal height is the simpler element, and that tlie depth of the u])per jaw is a disturbing element added to that in the whole uasi-alveolar height. We have already seen in curves 16 to 20 how hard it is hy inspection to distinguish what are the components when they are near together and of the same width. And therefore it might seem impossible, or merely fanciful, to distinguish the components in most of tlie curves here given. But by taking in one view tlie curves of quantities which may probably Ije related, such as the height and breadth of the nose or face, the interpretation of one curve will explain the others. For instance, in the first line, curve 32 shows that two groups are here mingled, and that tlie tall-faced group is rather fewer than the other. Two curves of similar relative areas are doubtless to be seen in No. 30, the length of the nose. Presumably also the longer face or nose is associated with greater width, and thus the two curves of the same relative areas will, when combined, give the curves 31, 33, 34. In the next line two curves were also distinguished in curve 35 ; and, after these had been traced in 36, 37, 38, lastly, on taking the biauricular width, 39, the lesser curve, was found standing almost entirely detached. Another application of this method may be illustrated, where it may seem (juite uncertain how a hummocky curve is to Ijc analysed. In 35 and 37, the two components are fairly clear. Taking now all the examples in one component that are beyond the range of the other, that is, all nasal heights below 50, and all over 54, and then plotting the two curves of the nasal breadths of these examples, we find that they have exactly the same centre, but one is much less spread than the other. These combined, exactly explain the form of curve 36. In another case, the bizygomatic width. No. 63, is the most certain evidence of the relative areas of the component curves in that line. Taking now all of the examples below 125, and all over 132, and plotting the curve of nasi-alveolar heights of these, we find the two curves shown in No. 62 as components. It is sometimes more practicable to distinguish components liy plotting a diagram as in No. 75, where the nasi-alveolar heights used in curve 32 (where they are easily distinguished) are crossed with the nasal heights of curve 30, which have to be disentangled. The diagram shows at once which of the nasal heights belong to each of the nasi-alveolar components, and enables us to resolve the curve 30 with accuracy. The second dimension gives a perspective to the view, which at once makes the two groups disentangle, though when foreshortened into one mass they are indistinct. lieduction to two curves of relative areas is not arbitrarily forced here ; in No. 39 is one exception which cannot be entirely reduced by two component curves like those in 35 to 38. The explanation is that the larger curve in oh to 38 is a compound of the earlier curves fused together, as we have noticed : and in the biauricular width they are not yet fused, but tlie separate curves of No. 34 are still showing in No. 39. There is no reason against PiiOF. W. M. FiJNDEKS Petrie. — Migrations. [39] three different stocks being traceable, out of the dozens really involved ; and though only two components are discussed here, that is because the material is insufficient for the precision requisite to disentangle a third factor. Doubtless each of the curves here are already composite in earlier stages, and partly fused into unity. The physical meaning of this tracing of curves of similar areas, hi one line of related dimensions, is that two different types of man are mixed together, l)ut not fused. The examples are probably merely mixed mechanically, all of comparatively pure parentage of one type or the other ; Ijut it may in other cases be due to atavism of one of the parental types, that is to say, that whatever causes bring forward a dimension of one type, may also bring forward other related dimensions of the same type. Probably the distinction between pure inheritance and atavism would be shown by unrelated dimensions being also found in pure inheritance, and only strictly related dimensions being found in atavism. But we have not enough material to discuss this question. The problem may perhaps be placed in general terms thus : — Given a series of curves of variation of correlated dimensions, to find two (or more) component probability curves, which when coml lined at varying distances, and of varying heights and In-eadths (their relative areas being always the same), shall be capable of forming all the compound curves in question. The practical method in this analysis is to look over all the correlated curves, and select whichever shows the components most separately, as it is only from such that the other curves can possibly be analysed. Draw the two outer sides of the curves which are clear of mixture ; note that the crossing point (c.P.) of the inner sides of the curves must be at half the height of the compound over it, and the angles of the slopes through the c.P. must be equal and opposite to the angles of the other sides of the two curves. By inspecting the types of the compounds given here, curves 11 to 29, and while remembering the various figures of the probaljility curve of different heights (as shown also in 11 to 29), a trial should be made of completing the best defined curve symmetrically through the G.P. Then the height of the compound curve above it at each point should be brought down to tlie base line, and the second proljal)ility curve plotted through these heights. If it fairly conforms, this may be accepted ; or else the first curve may be widened or narrowed to alter the material left over for the second curve, until two curves can be found which shall yield the compound curve as closely as may lie. But tlie irreducible requirements are: (1) that every curve shall be a probability curve ; (2) that every curve shall be equal sided or symmetrical ; (3) that the C.P. must be at half the height of the compound ; (4) that the area of overlapping of the curves below the c.P. shall equal the area included in the compound over the c.P. ; (5) that these two areas shall extend to the same distance either way ; and (6) shall be as nearly as possible of the same height at any vertical line. Before leaving this we should notice S(jme cautions and limitations; and the bearing of other researches on the position here taken. First we must not assume [40] PuoF. W. ]\I. FuNDEKs rETiiiE. — Migrations. that the proceeding liere is applicable to a fused race, but only to a mixture of two races. Of course, when two races live together, at first there is mere mixture, at last there is fusion ; and the actual results wliich we study are tliose of the combined curves of the two elements mixed, smudged in all directions by the polymorphic variations let loose by the disturbance of fusion. But still the mixture is all that we can profess to analyse. In fusion there will no doubt be many examples of reversion : that is to say, in the Mendelian distribution of variations the extreme cases may be conveniently labelled for our purposes as atavism. And atavism may appear in one element or more. We need practical studies as to the proportion of atavism, and its correlation in different dimensions, both in those that are related as nasal height and width, or nasal height and nasi-alveolar lieiglit, and in those that are unrelated. The disturbance due to fusion of r//scontinuous variation, as in different colours of animals, or horned and unhorned breeds, is not yet shown to l.>e similar in continuous variation, such as the few per cent, greater growth of one bone, which may vary without a stress between different types. Simple fusion variation seems much more likely to occur where the difference is only a minute chancre in the amount of a single activity. The whole question of skew curves of variation, instead of symmetrical Gaussian curves here adopted, is not yet clear. It has never been shoM'u that a truly single, natural, inorganic cause does not vary in a symmetrical manner. In saying tliis it must be remembered that variation is on a logarithmic scale and not by natural increments, as I pointed out in " Pyramids and Temples of Gizeli," and illustrated by an extreme case of conscience money in " lieligion and Conscience." Hence when skew curves appear in organic variation, it is at least the first hypothesis to try if they may not be the result of an unlimited complex of compound curves, due to tlie infinite complexity of conditions and ancestry. "Whetlier a test can be devised which will exclude such an unlimited complexity as a sutficient cause for the skewness of any given set of observations, seems extremely doubtful. At least it has not yet been done so far as I am aware. We must all respect mathematical analysis as the elaljorate tool for doing what nothing else can do ; but its efficiency all depends on the Ciuestions which it is set to solve, and tlie framing of those questions must be done with due regard to all the causes, which can only be distinguished in the results where thev have been anticipated in tlie framing of the questions to be solved. Before parting from the curves wliicli we have here attempted to analyse, it should be noted that the possibility of so doing varies greatly in different groups. In the early prehistoric, the nasi-alveolar height (32) gives a clear separation, and shows what relative size of groups may be expected in this age. The late prehistoric is well divided by the biauricular width (39), tliough that was not noticed until the others had been analysed. The I-II Dynasty is divided best by the nasal width (41) but this is not so certain as the division in other ages. The III-V Dynasty Prof. W. M. Flinders Pejtje. — Migrations. [41] is well divided by the biauricular width (49). The VI-XII Dynasty is uncertain in division, but that shown seems the more probable ; the curves are too near together to be safely separated. The XII-XV Dynasty is tolerably separated by the bizygomatic width (58). The XVIII Dynasty is again divided by the bizygomatic (63), and somewhat by the nosal width (61). The Ptolemaic is not clearly resolu- able ; the nasi-alveolar height (67) is the best indication. The Eoman groups are very uncertain in result ; thus we see the great use of looking at many different measurements, as where four out of five curves may be insoluble owing to having the components too near together, the fifth may give a good analysis, as in such cases as curves 32, 39, 49, or 63. This process must not, then, be judged as if it made a vain profession of resolving the majority of curves apart from others ; its possibilities entirely depend on searching so many dimensions that one may be found to help to a solution of the less distinct curves. D. TJie conclusio/ift to he drawn. Having thus resolved these compound curves of one period each into two separate curves, we reach the practical question of the interpretation of the results historically. Looking at curves 30 to 34 it is seen that in tlie early prehistoric age of Egypt, say 8000-7000 B.C., there were two types ; one which was rather more numerous, had a small face, about 122^ bizygonuitic width, and only 66 in nasi- alveolar height, and with a short nose 48 high and 27 wide. This is as short ])ut not so broad as the the negro type, which is about 46 to 50 high and 27 to 30 wide in different groups. There are other short-nosed types nearer Egypt, as in Sinai, where an entirely un-Semitic and Socratic type may be seen. The other type of this age had a larger face, of 129 J bizygomatic width and 73 in nasi-alveolar height, with a longer nose of 52.^, and narrower, being 24|. This nose is the Algerian size, and of the usual European proportion. So far we may say that there was a lower race intermediate between the Algerian and a Socratic or negroid type, and a higher race comparable with Algerian or European. Though there is a difference of l/20th in the breadth of the face, the average capacity only differs by l/12th, showing that the other dimensions of the head were not smaller in proportion. In the later prehistoric age, say 7000-5500 B.C., there was in each dimension one large group which occupies the mean place of the two previous groups. This is markedly the case in all but the nasal width, where the difference of position of the groups is too small to show this. The large group seems to be the result of fusion of the preceding groups. Beside this there is a small group with longer nose and larger face, altogether a higher type, slightly better in each dimension than the higher group of earlier times, and not probably developed from that, or else it would merge in the large group. The nasal height of 53^ and breadth of 26 might well belong to a good type of Semite or European race. So far we have dealt with the material from Abydos, El Amrah, and Hou, a district of about twenty-five miles. But forty miles beyond that, further up tlie [42] Vuov. ^y. M. Flindeus TETitiE. — Mifjmtions. Xile, is the Xaqada district, from which a large series of skulls was brought, now published in Biometriha, i, pp. 408-67. As the nasi-alveolar height gives the distinctive curve of two elements, No. 39, this was extracted from the Naqada measurements, sorted into male and female, and also eaily and late by the sequence dates of the graves. Owing to this series having been found long before plates of the pottery were drawn or sequence dates were invented, the dating has to depend on the registration of the pottery which was preserved ; hence where there was no good or perfect pottery the dating could not be recovered. In the later excavations, after plates of pottery types were published, it was usual to carry a set of type plates of pottery to the work, and register every form, no matter how broken. Thus the dating is far more complete for the skulls from later excavations (published by Mr. Maclver) than it is for those from earlier excavations published in Biomdrilri. To get enough to show any result evenly, it was necessary to include both male and female skulls. In the early prehistoric tliere is no difference in the median, both being 68 ; but in the late prehistoric the male median is 67, and the female Go. Hence the latter were all raised by 2 units to 67 in order to be added to the male skulls. In order to avoid any question about comparing mixed sexes in this way I liave similarly treated the female skulls from the Abydos region. In curve 76 are the female skulls alone. They have a median of G7i, which is 1 below tlie male median of 68^. Hence they are raised h\ 1 unit and added to the male skulls in the curve 77. This bi-sexual curve from the Abydos region shows the deep cleft in two parts very distinctly, as in the male curve (32) alone ; and the female curve alone (76) shows tlie same cleft, but not so prominently. Xow, turning to the Xaqada early prehistoric curve 78, we see it is of a totally different shape from 77. There is no second element at all comparable with that seen in 77, and for Xo. 78 a single curve with an axis at 68^ seems all that can be fixed. There is no trace of the great peak at 73, so marked in curve 32. It seems, then, that there is a local difference, and in the Abydos district there is a considerable long-faced element which is not found at Xaqada, further up the valley. "Which of these two types most pervades the whole country is quite unknown. Of the late prehistoric age the Xaqada curve 79 appears to be single, and to ha"\e little or none of the higher element which was fused into the larger curve in No. 37. Taking the numbers, 66 4- 73 produce 69 at Abydos ; while at Xaqada 68 has slightly retrograded to 67, not having any addition to raise it. Moreover, there is no trace at Xaqada of the new high element at 73 which is so distinct at Abydos. Here again the Xaqada results show less mixture than those lower down the river. Coming to the beginning of the kingdom, we see in Nos. 40 to 44 two equal curves. These are rather nearer together (in all but nasal height) than the large and small curve ot: the previous age. It might well be that the high type minority had increased, and some fusion had occurred between that and the lower majority Some new inHuence is, however, suggested by a separation of the two types of nasal breadth. Pkof. W. M. Flinders Petrie. — Migrations. . [43] In the pyramid age of the III-V Dynasties no fresh type is seen ; the face became rather higher and narrower in both types, but otherwise there is little change. By the break up of the old kingdom a change conies in. One large curve occupies the place of the two seen before, and the races were fused to a medium type. lUit by the side of this there is a large addition of about half as much again of a lower type, with narrower and shorter face, and shorter nose, while the width of the nose remains the same and is thus broader in proportion. This race is almost exactly the same in face and nose as the lower type of the early prehistoric. Passing to the great age of the Xll Dynasty and succeeding centuries we lind again a group intermediate between the two of the previous age, apparently another fusion of the race. But by its side is as large a group of a higher type, apparently a new stock about equal to the higher group that came into the later prehistoric age. The XVIII Dynasty continues, apparently, these two types, with perhaps a very slight approximation together. On reaching Greek times we again see in each dimension a large fusion group nearly in the middle of the two preceding groups. And added to this is a new group of lo^^'er type with short face and no.se, and narrower face, just as low in each respect as the lower of the prehistoric. In Iioman times the same groups continue, but the lower group is as numerous as the better type. The summary of the changes that we can trace in the facial dimensions, then, are as follows : — Early Predyiiastic .... Low type + High type. Late „ ..... Fused type + High new. J ' i I -II Dynasty ...... continued. increased. III-V Dynasty slight improvement in both. V ^ / VI-XII Dynasty .... Low new + Fused type. ^ ,_, ' XII-XV Dynasty Fused type + Higli new. XVIII Dynasty Slightly nearer Ptolemaic .... Low new + Fused type. Roman ..... Low increased. continued. Thus the great changes have been a high and a low type alternately introduced twice over. These results must only l^e looked on as a first attempt on new lines, where all previous work has not yet cleared the subject. It is to be hoped that a much wider examination of all the dimensions, on a larger body of material, will enable such a research to be carried out far more completely than is possible as a bye-issue in the midst of other work. The various questions about the variability of fused races also need much more illustration, before we can interpret what we here see with certainty. [4-4] Prof. "\V. M. Flinders Petrie. — Migrcdions. Explanation of Plates XXVII-XXIX. XXVII. Curves 1 to 10 show the diminution of irregulai-ities by the increase of material, 1, 2, formed fi'om 167 examples ; 3, 4, show three curves, each formed from 50 examples of the same measures as 1 and 2, the variations between the curves showing what are non-significant irregularities ; 5, 6 show curves from 30 examples ; 7, 8, from 20 examples ; 9, 10, from 10 examples. The difference between the columns is that in the first column, 1, 3, 5, 7, 9, there ai-e 10 divisions in the .scale ; in the second column, 2, 4, 6, 8, 10, there aie only 5 divisions. The avoidance of cjisual iiregularity by thus increasing the number of examples per division is clear. Curves 11 to 1.3 show the compounds of two curves where one is of variable height, and the distance is sufficient to show the components ; 16 to 20 show the same components nearer together when they can scarcely be distinguished in the compound ; 21 to 25 show the same cui'ves so close together that they aie indistinguishable in the compound. The vertices of the cui'ves, rn- medians are marked with a short line ; and another line is put at the probable error limit, or quartile, on the right hand of the major component. Curves 26 to 28 show the compounds of two curves of equal height, one half the width of the other, and distance variable ; 29 is a near approach to a flat-topped curve. The use of all these forms is to .show what components may be expected in various forms of compound curves. XXTIII-IX. The curves 30 to 74 are thtise of five dimensions of male skuUs of successive age.s. The full line shows the total of the measurements ; the dotted curves within it show the probable components. These components have their relative size indicated by the curves with most marked division, as 32, 39, 49, 58, 63. The solution of the other compound curves by two curves, which shall have the same relati\'e areas along the line of one period, shows what are probably the components. Diagram 75 shows how a group of nasal height which was insoluble, could be truly divided into two curves by crossing it in a diagram with the nasi-alveolar height, which was already clearly divided by its curve, No. 32. Curves 76-79 are to show the difference between Abydos and Naqada skulls. The male and female had to be treated together, as there was not enough dated material from Naqada to give a result for one sex only. The curve 76 is the female curve, the male being No. 32 above. This (76) was raised 1 mm. to accord with the mean of the male skulls and then added with that to form curve 77. This is comparable with 78 and 79, in which it will be seen that the highei- group at 72 mm. is absent. Cun-es 80 and 81 show the maximum length, for male and female added, in order to be comparable with the skulls of mixed sex from Algeria, 82 ; the Ptolemaic group 83 is also closely iu accord, and is given as it is probably the nearest in date to the Algerian. Curves 84 to 87 show the maximum width, of the same classes as those of the length. The signs i 5 show the position of the averages of male and female separately. M is the median of the Algerian. {^Reprinted from the Journal of the Anthropological Institute, ToJ. XXXFI.^ Harriton and Sons, Printers in Ordinary to His Majesty , St. Martin's Lane, London. Journal of the Anthropological Institute, Vol. XXXVI., 1906, Plate XXVII. 1 ^2 ^26 VARIED AMOUNTS OF MATERIAL. ^■■II^^HI ^^^^^^H ^^^^^|B^^^^^^^^^^ ^^^^^^^^B^ T^^^H ^^^^^IBPI^^^^^^^^ ^^^^^^^^^v ^^^^m^^^^^^^^^^^ ^^^^^^^^^v ^^^^^^1 ^^^^^^^^^^^^^^^^^ X^^^^^^^IP ^v^^^^H ^^^^^^^^^^^^^^^^ff ^^pi ^l^^^^l Bale Si DsnieNson, Ltd., Pbato-Uth VARIABILITY OP CURVES, AND FORMS OP COMPOUND CURVES. Journal of tlie Antliropological InstiluU, Vol. XXXVI., 1906, Plate XXVIII. 50 55 20 2'i 30 60 65 ,, 70 75 80 EARLY PREHISTORIC ".. NASAL HEIGHT & WIDTH NASI -ALVEOLAR HEIGHT 75 EARLY I 50 PREHISTORIC -I 70 EARLY PREHISTORIC NASI- ALVEOLAR HEIGHT NASI-ALVEOLAR HEIGHT rJ^' ABYDOS. I 7 VVl FEMALE K i \ ABYDOS i FEMALE CORRECTED ; , & MALE ii_ r^ ; * 77 NAQADA EARLY NAQADA LATE # M r\. Bale k Da«ieluoii, Ltd., Ffaeto>litJ CURVES OF EGYPTIAN SKULL DIMENSIONS. Journal of the Anthropological Institute, Vol. XXXVI., 1906, Plate XXIX. BIZYGOMATIC WIDTH Bl AURICULAR WIDTH MAXIMUM 170 LENGTH EARLY PREHiSTORlC 81 LATE PREHISTORIC ALGERIAN DOLMENS ROKNI A 140' MAXIMUM WIDTH PTOLEMAIC CURVES OF EGYPTIAN SKULL DIMENSIONS. Bale & Dsnielsson, Ltd., Photo-lita ANTHROPOLOGICAL INSTITUTE OF GREAT BRITAIN AND IRELAND. This Institute was established in 1871 by the amalgamation of the older Anthropological Society and Ethnological Society. It hsis for its object the promotion of the Science of Mankind^Antliropology — by the accumulation of observations bearing on man's past history and present state in all parts of the globe. It investigates, in a rigorously exact manner, everything that can throw hght upon the laws of man's nature : his origin, history, and capabilities of progressive development and civilisation. It studies him structurally and psychologically, under the several types in which he is found in various regions ; and comparatively, in relation to the rest of the animal kingdom. The ultimate object of the Institute, therefore, is to build up a Science of Man on a basis of well-ascertained fact, and logically deduced inference. As means to this end, the Council of the Institute has adopted a plan of operations, the principal heads of which are as follow : — 1. Meetings for the reading of papers and for discussion of anthropological questions. 2. The issue of a Journal containing Eeports of the Proceedings at the Meetings, with other matters of anthi'opological interest. 3. The appointment of Local Correspondents in all parts of the world to collect information, and to aid the Institute in its operations. 4. The maintenance of a library, which contains sets of all the principal Anthropological Journals published in the United Kingdom or elsewhere, besides a large and valuable collection of books of reference, travels and researches, an extensive collection of photo- graphs and lantern slides, and much unpublished material for the use of students of Anthropology. Fellows residing in the country, as well as in London, can borrow books from the library. 5. The appointment of Committees to conduct special investigations, as occasion offers, in the various branches of- Anthropology. 6. Co-operation with the British Association for the Advancement of Science, and with foreign scientific societies in anthropological investigations ; co-operation with individuals and institutions in aid of explorations and in the establishment of local centres of •anthropological study ; and, generally, the stimulation of individual and local efiforts to further the objects of the Institute. The annual subscription is Two Guineas, which is due on election (unless such election takes place in the month of November or December) and on the Ist of January in each succeeding year. A Member may at any time compound his future subscriptions by the payment of £21. Persons who wish to become Fellows of the Institute are requested to communicate with the Secretary, 3, Hanover Square, W.