QUICHUA AND AMERICAN AFFINITIES AS DEDUCED FROM NUMERALS.
This investigation may be opened with the numerals, r two', 'four, and 'eight', which are, in Quichua, yskay, ttahua (-/ma=English -wa), and pussak. This last is written by Markham pussac, and by Nodal puzac. ' Four* is given by Nodal as tahua, by Markham as ttahua, and by Lopez as i? tahua ; and there are some other slight differences of orthography in other terms, but none of consequence. The Quichua 'four', ttahua, is compared by Lopez with the Aryan dva-dva, ' two-two*; and yskay, two', with such a form as the German zwei. I should here go further than Lopez, as I believe that not only the Quichua and the Aryan 'two' and * four', but ' eight' as well, are ultimately identical in both systems of language; and that, in each, ' four* =' two-two', and e eight' =' two-two-two'. Yet, in spite of this, I cannot coincide in the opinion entertained by Lopez, that there is a peculiar affinity between the Quichua and the Aryan, to the exclusion of other languages : for I should consider that, as far as these three numerals are concerned, the Quichua is identical, not with the Aryan alone, but also with a num- ber of languages in other parts of America, as well as with the Basque, the Georgian, the Circassian, the Tungusian, the Chinese, the Siamese, the Malay, and the Polynesian. In comparing languages together, or at least in comparing numerals, which commonly preserve the earliest rudiments of a language, any induction that is to be sound and impar- tial must be extended over the widest area possible. The evidence of the ultimate affinity of all languages may other- wise be mistaken for evidence of a special affinity between two languages in particular, each of which has nearer rela-
PERUV1A SCYTHICA. 15 tives elsewhere. Etruscan researches frequently err in this respect. If the view given above be correct, that the several 'twos', 'fours', and 'eights' mentioned have a common origin, then there would be in all three numerals only a single element, r two\ What the complete form of this 'two' originally was, is uncertain; but it seems to me to have been not far from gwar or agwar, and to be resolvable into 1 + 1, ga (or atf)+war. Gesenius, however, judging from the Semitic and Aryan languages, both of which possess the 'two' in question, has inferred the original form to be rather different. " The primary form of the numeral", he says, (Heb. Lex., ed. Tregelles), " appears to be *Jft, from which have been softened Sanscr. dwi, dual dwau, compare twa, other, different, Goth, twa, twd, twai, whence Eng., Germ. two, zwo, Gr., Lat. 8vo, duo." According to this supposition, it should be an n that our own ' two' has lost, instead of an r, as implied in the supposition that the original form was gwar. But it would, in either case, be some liquid that is deficient ; and this may be confirmed from the Tungusian languages : for they give us ?wr, $ur, and dyul, for ' two', as well as du-in, dy-gyn, and similar forms, for 'four*, =' two- two'. In America, the liquid appears, not as r or Z, but as n ; as, e.g., in the Galifornian ' twos', pen, eshin, and teene, which last correspond closely with Semitic ' twos': but in Aryan ' fours', such as the Gothic fi-dvor, etc., and in the Basque and Georgian ' eights', zor-tzi and r-va, the liquid is an r, as also in the Lazic 'two', zur ; a West-Caucasian numeral whose elements, each implying 'one', may be found in the East-Caucasian ' ones', sa, zo, za, and in the Lazic ar and the Georgian erthi, c one'. The suffix in the Georgian er-thi, 'one', which may be compared with the Basque er-hi, 'finger', is seen, reduplicated, in the Georgian thi-thi, 'finger' (cf. Chinese ii> 'finger, toe'), and probably in the Armenian tharth, ' hand', = ' fingers', =' finger-finger'.
16 PERUVIA SCYTHICA. In the Quichua yskay, ' two', I take -kay not to be a part of the original ' two', as -ei would be in the German zwet (with which Lopez compares yskay), and in the Mantshu dzheio or tfeio; but rather to be a mere suffix, like -cua, -cu, -ko, -he, -k, in other American numerals; -gu, -go, -ba, -khi, -thi, *i, -kh, -t, in the Caucasus ; and -gu, -ko, -ku, -ya, -ia, in the Himalaya and India. In the Quichua kand-is, 'seven', a numeral commonly equivalent to ' five-two', the supposed suffix -kay , in ys-kay, ' two', would not appear in the final element of 'seven', -is, 'two'. This would reduce 'two* in Quichua to ys or is, or even to *, if y or i be only a prosthetic vowel, as it would be if the original form of 'two* was tfwar, instead of atfwar. But this is unimportant. Of the 'eights' comprised in the list soon about to be given, two are clearly enough = 'two-four': for the Ende (Malay) rua-butu, ' eight', is manifestly = Ende rua-wutu, 'two-four'; and the Netela (California) weheswaza, 'eight', would in like manner be = Netela wehe-waza, 'two-four'. But each of these ' eights' I should be inclined to take as reconstructions ; as comparatively modern formations from ' two-four' to supply the place of the ancient ' eight', which had ceased through phonetic decay to indicate the factors of which it was composed, and had thus lost some of that significance which would tend to facilitate computation, and make its results clearer to a barbarous mind. Yet we can hardly doubt that these ancient ' eights', thus supposed to be replaced, were originally formed in the same manner as their modern substitutes ; for one of them, the wehe-swa- za (cf. German zwei-zwei-zwei) of the Californian Netela, seems only a more perfect form of the St. Baphael (also Californian) wu*su-ya, ' eight', or of the Kutani ' eight', war wahsah, in the mountains of Oregon, which last form, wawahsah, is still nearer to the Netela wehe-waza, 'two- four', than it is to the Netela weheswaza, 'eight'. It is obvious that the St. Baphael 'eight', wusuya, is an old
PERUV1A SCYTHICA. 17 form, and no modern construction from the St. Raphael oza, 'two', and wiag, 'four': and a similar conclusion re- sults from comparing the Kutani wawahsah, ' eight', with the Kutani ass, 'two', and hazah, 'four\ The other of the two supposed recent ' eights', the Ende (Malay) rua-bu-tu, is indeed more remote from other Malay ' eights', such as wa-ru and fa-lu, than the American ' eightsf just cited are from one another ; but the variation is not so great as to prevent our discerning that such other Malay ' eights' were originally ' two-four' or 'four-two', both ultimately = 'two- two-two\ This, indeed, is plain enough from Polynesian ' twos', 'fours', and ' eights', like the New Zealand rua, wa, wa-ru, or the Marquesus ua, ha, va-u. And here we may observe, before we pass on, in illustra- tion of the superior antiquity of the American to the Poly- nesian languages, how much more perfect than such Poly- nesian forms as these last, are a pair of kindred 'twos', 'fours', and 'eights', in Central America: - the Gunacuna vocu&, pa-kecua,, vo-u-agua,; and the Bayano ^ovuar, pa- ievuar, pa-va-ke. As far as 'four' and 'eight' are con- cerned, the Quichua tta-hua and pu-ssa-k might be adduced in confirmation of the same view. Pake and ttahua, 'four', are completer than the Polynesian wa and ha; and vatia, pavake, and pussak, ' eight', than the Polynesian van and warn. But, when we fall back from Polynesia towards the west, the balance is restored, at least as far as regards 'four', as well as ' two', though American ' eights' still retain the advantage in point of completeness over Malay and other 'eights'. In the following examples from the Papuan area, which I have taken from Latham (Comp. Phil., p. 381), America and Oceania stand nearly on the same level of completeness in ' two' and ' four': ' eight' is omitted, be- cause it is composed in these Papuan vocabularies of ' five- three', instead of 'two-two-two'; and affixes are left un- italicised : - D
18
PERUVIA SCYTHICA.
' Two' . ' Four', = ' t wo-two'. American. Quichua i/akay tta-hua Bayano 2>0vuar pa-kevutiv Cunacuna vocua pa-kecua Papuan. Brierly Island pahiwo paihe/>a-£ New Caledonia on&dou ou&tbai-t For the affixes here, see my Numerals (p. 44), in which I have traced the presence of similar and probably kindred forms in the Caucasus, in Nepal, in India, in China, and in America. We may now proceed to the full comparison of the three numerals, ' two', c four', and ( eight', where such affixes will be again left unitalicised. Two, fowr, eight.
Patagonian yjene-ke-kague 1 ke-kagny xeuksy That wene- here is « two', appears from the Patagonian wene-cash,
JPERUVIA SCYTHICA.
19
< Eighty = North America. 2x2x2. Cunacuna (Darien) va-u-agu& Bayano (Guatemala) pa-va-ke Pima (New Mexico) ki-ki-g Tesuque (New Mexico) kuh-beh Attakapa (Texas) zi-kui-au Kioway (Texas) ia-za Chemehuevi (New 1 Mexico) j Netela (California) weJieswa-za St. Raphael (California) wusu-ya Talatui (California) ka-wm-da Pujuni (California) pe-(fe-i Sekumne (California) ta-p-wi St. Barbara (California) ma-la-hua St. Luis (California) shko-mo Kutani (N.E. Oregon) wa-wah-sah Unalasbka (Aleutian) . . . Old World. Yeniseian Kanskoi(Upper Yenisei) shi-dd-td Samoyed a-ghe-8 Tshuvash (Russia) sa-ka-r Turkish se-ki-s a-gy-s
with the Calif ornian 4 twos', below, wche, pen, teene, and eshin; also with the first element in the Tesuque ion-ouh, 4 four*, 2x2, and with the middle element in the Talatui ka-wm-da, 4 eight', 2x2x2. 1 Perhaps -koho here is only a suffix ; but it has passed into the St. Barbara s^u-mu, 'four 1 , and the St. Luis sh/fco-mo, 'eight'. The intru- sion of an m into these two numerals, and into the St. Barbara * eight', ma-la-hua, is rather singular; but a solution of the apparent anomaly will offer itself directly.
Chinese pa-t sze « i pah sze ar, eul Siamese pe-t si son Gyarung (China-Tibet) or-ye-t kadi 1
Kol (India) i-r-Zia u-purim barm Khond (India) i-l-arku. u-pariku barku Burmese . . . le
Sunwar (Nepal) zoh le
Basque (Pyrenees) zo-r-tzi la-u bi Circassian (Caucasus) yi t-ley oh Georgian (Caucasus) r-va o-thkhi ori Lazic (Caucasus) o-vr-o o-tkh zur Etruscan dice (Tuscany) . . . hu-th thu Tuschi (Caucasus) . . . dhe-w shi? Lesgi (Caucasus) bei-t-lgu bo-ogu degu mi-t-lgp 3 u-kgo kigo
1 Compare the Gyarung prefix, £#-, with the Quichua suffixes, -kay and -ka, and other suffixes in America, the Himalaya, India, and the Caucasus (ante, p. 16). 2 The base is shin (Schiefner), which is like the Hebrew then-. Three other forms of the Tuschi 'four' are given by Schiefner: - bhew (cf. Lesgi booga), and also whew and yhew; between which and the Georgian othkhi the huth of the Etruscan dice is intermediate. In the Basque 4 four', lau, the d of the Tuschi dhew, 'four', is weakened into I, as it is still further, in the Tuschi yhew, into y. 8 The change of bei- into mi- here, and the corresponding change in the Ude ww-gh, * eight', immediately following, may explain the appear-
PEBCVtA SCYTHICA.
21
' Eight'^ 'Four*,= ' Two', Old World. 2x2x2. 2x2. 2. Ude (Caucasus) mu-gh bi-p l pha Abkhazian (Caucasus) a-aba ph-shibb tnba Gothic a-h-tau firdv&r tvai Welsh wy-th pe-dwar dwy English ei-gh-t f-our two Armenian ov-th d-or erku Latin o-c-to qua-tuor duo Wallachian o-p-t pa-tm doi Russian vo-sem' (fe-ture dva Sanskrit ash-tan da-tvar dm, dva Hebrew . . . ar-ba 3 shenaim Syriac . . ar-bo' theren Amharic (Abyssinia) . . . . arr-u
ance of an m in some Californian numerals, such as the St. Barbara ma-la-hua, ' eight' (ante, p. 19), with which compare the Lesgi mi-t-lgo, 'eight'. In ma-la-hua, ma=dva=htca=dva=la. 1 Bi-p is a doubling of the Basque &», 'two'; a numeral which supplies one element to the Basque zaz-jpt, 'seven', 5 + 2, and to the Abkhasian fo'-sh, 'seven', 2 + 5 (ante, p. 10). This Basque bi, again, when com- bined with the Circassian tley, 'four 1 , would produce the Lesgi (Andi) bei-tlgu, ' eight', 2x4. The Basque 'two', ' four', and 'eight', with the 'two' and 'four' of the Etruscan dice, do not differ more from Caucasian 'twos', 'fours', and 'eights', than such Caucasian numerals differ from one another. All might be classed together under the name "Iberian". 9 Compare this African 'eight', tag was, with the Turkish 'eights', sekis and agys, and with the Samoyed ' eight', aghes.
22 PERUVIA SCYTHICA.
' Eight*,= ' Four\= ' Two', Old Would. 2x2x2. 2x2. 2. Malay- Kayan sa-ya pa-t duo Bima wa-ru opa-t lua New Ireland wa-l ha-t ru Eotti/a-/w ha-a dua Pelew ta-i o-ari oru Caroline wa-l tan ru Fiji wa-lu va rua Hawaii va-lu ha lua New Zealand wa-ru wa rua Marquesas va-u ha ua
In the preceding table, some languages agree with the Quichua in one of the three numerals, 'two', 'four', and ' eight'; some in two of those numerals ; and some in all three. The list of languages connected by such a triple concord extends to these families or countries : - Patagonian. Quichua. Darien. Guatemala. New Mexico. Texas. California. Oregon. Tungusian. Samoyed. Turkish (if ' two' may bo included). Chinese. Siamese. Malay o-Poly nesian . Sudan. Basque. Ude.
PBRDVIA SCYTHICA. 23 Lesgi. Georgian. Circassian. Abkhasian. Aryan. These languages may be collected into five classes : the American, the Aryan, the Turanian, the Iberian, and the Sudanian in Africa ; the Sudanians being the members of the African race who would have remained longest in con- tact and communication with other races than the Semitic, and the Sub-Semitic or Libyan. As, then, one of these five classes, the Aryan, has been deduced from Mount Imaus, so it may appear probable that the plateau of Central Asia, of which that mountain-range is the western rampart, was the region where the previous s twos', ' fours', and ' eights' first took form, and whence they spread over so vast a portion of the world. Supposing, therefore, that plateau to be the place of their origin, those three numerals would have been carried on one side towards the south and west, first by the Sudanian Africans, then by the Iberians, and, finally, by the Aryans ; and, on the other side towards the east, first by the Americans, and afterwards by the Tura- nians. For we may link geographically together, in the following manner, the nations and countries, just enume- rated, which have c two', i four', and r eight' in common : - Patagonia Peru Darien Guatemala California - New Mexico - Texas Oregon Tungusian \ Samoyed > Turanian. Turkish 3
► America.
24
PERCVIA SCYTHICA.
Plateau of Central Asia. Aryan Kol ^ Abkhasian ^ Chinese
Circassian Georgian Lazic Lesgi Ude Basque Sudan
Iborian.
Siamese Malay Polynesian v
Turanian.
J
That the connection between these languages is one which dates from a very remote antiquity, appears not merely from the length of the chain formed by them, which reaches from Sudan to Patagonia, but also from the fact that it is frequently difficult, if not impossible, to take the ' two' of any particular language among them, and from it to derive the ' four 9 and ' eight' of the same language, the one as 'two-two', and the other as ' two-two-two'. Our own two, four, and eight, may be taken as examples. Here we must resort to other languages, and those languages not even Aryan, in order to make out the derivation ; although, when that derivation has been made out, we may see, in such an instance as that of the English ei-gh-t, or of the German a-ck-t, how three 'twos', like those which would be required to form these 'eights', might be collected from three Aryan 'twenties', the Greek eikosi, the Armenian cftsan, and the Swedish ljugo. The case is similar in other languages, always excepting such instances as I have noticed above, and considered as comparatively new reconstructions of older ' eights': the Californian weheswaza, = wehe-waza, ' two-four' ; and the Malay ruabutu, = rua-wutu, ' two- four'. We must, as it appears to me, suppose that a number of 'twos', 'fours', and 'eights 1 , our own three numerals included, were constructed and became current in
PERUVIA SCYTHICA. 25 Central Asia before the progenitors of the Aryans, the Iberians, the Turanians, and the Americans, ceased to be in communication with one another; and that each race appropriated to itself, out of the floating capital of variants, the oldest forms of those which their members now employ. The following numerals, selected from the list given above, may be used to show in what manner different and remote languages serve here to explain one another : - Polynesian ha, ' four* (one ' two* deficient) . Malay haa, ' four*. hat, 'four'. pat, 'four'. Chinese pat, 'eight' (one 'two* deficient). Siamese pet, ' eight*. Californian pedei, ' eight'. Peruvian (Quichua) pussalc, ' eight'. (Aymara) pusi-pa, ' four-two'. Abkhasian pAa/iiba-viba, ' four-two*. aaba, ' eight' (one ' two' deficient) . Anglo-Saxon eahta, ' eight'. Armenian ovth(=uth),' eight' (one 'two' deficient). Etruscan dice hut h, ' four'. thu, 'two'. Circassian oh, 'two'. Georgian othkhi, 'four'. Lesgi ukgo, ' four'. Latin octo, ' eight'. Malay wutu, ' four*. Welsh wyth, ' eight' (one ' two' deficient). Tshuvash dwatta, ' four'. Mantshu tuye, ' four'. Peruvian (Quichua) ttahua, 'four'. Pelew tax, 'eight' (one 'two' deficient). Californian tapwi, 'eight'.
20
PERUVIA SCTTHICA.
Upon the whole, in the Old World, the Chinese and the Siamese languages, allowing for their having lost much in their numerals Ln phonetic decay or noJe other cause, appear to come nearest to the Quichua in ' two', ' four*, and 'eight', when taken collectively. The affinities may be thus presented : -
Shan tet ze zoun Siamese pet 81 son 1 Chinese pat sze v> Quichua pussak ttahua ysksij The monosyllabic character of the Chinese language may have operated to reduce these numerals to monosyllables. One American language, the Otomi of Mexico, has been, on
1 Compare the Chinese shwdft, ' pair, brace, couple'. * Our 4 two 7 is more nearly found in the Chinese Uoi, tsdi, ' a second time, bi$\ Cf. Japanese tsui, l pair\
PBRUVIA SCYTHTCA. 27 account of its monosyllabic character, specially connected by some with the Chinese, and separated from the other American languages ; a separation which has been rightly disallowed, and which seems as inadmissible as that of the Quichua from the same family of languages. 1 Other American languages are quasi-monosyllabic, one of them being the Attakapa, a language of Texas in the list imme- diately above. 2 It is, as may be discerned here and else- where, the languages of California, New Mexico, and Texas, which afford the nearest and most numerous North American parallels to the Quichua; an affinity tending to connect the Quichua nation with the Pueblo Indians, a people who, when first discovered by the Spaniards, were only inferior in civilisation to the Mexicans and the Peru- vians. 3 Before passing on to the next members of the Quichua decade, another illustration may be given of the fact, that Chinese numerals may be considered as nothing more than a reduction to monosyllables of numerals existing more per- fectly in America. An Iberian ' hundred', and some Iberian ' eights', seem to have suffered similar mutilation : - ' Hundred'. ' Bight'. ' Four'. ' Two'.
American - Quichua padah pussak ttahua
Aymara . . . pusi pa Turanian - Chinese pak pat eze m I peh pah
pai pa
Iberian - Tuschi bah . dhew 8hi Abkhasian . . aa phshi vi Ude . . mugh bip pha Circassian . . . yi tley oh
1 Latham, Comp. Phil., p. 431. « lb, * Domenech, The Great Deserts of North America, ii, 52. Domenech is inclined to regard the Pueblo Indians as descendants of some Toltee
28
PEEUVU SCYTHICA.
Here the Chinese pah bears to the Quichna padak an analogy like that which the English four bears to the Gothic fidvor, or the French huit and the Italian otto to the Latin octo. One, three. The next Quichna numerals which I propose to take in hand are 'three* and 'one', kimsa and huh; kimsa (in Cayuvava, kim-isa), 'three', being resolvable into 'one-two*, and the second element in kim-sa or kim-isa being, like the second element in pu-*sa-k, 'eight', equivalent to ys-ksy, ' two', and the kindred terms already considered. We shall thus have to find parallels for kimsa, ' three', Mm-, ' one', and huk, ' one': -
colonists "who settled on the banks of the Gila and the Rio Grande long before the Aztec emigration to Mexico". One Pueblo language is the Tesuque, which will often be cited in these pages. 1 Here -ga and -k would be suffixes, like those which have been already so frequently noticed.
kuma ken kain koen nioet ybani paini pin kan, kani kon gun kam
. . . huda . . . hogun ... C%iC dun dik dit The closest Quichua affinities seem here to be American, and next Australian; both Americans and Australians haying parallels on one side with the Turanians, and on the other with the Africans : or, if we class races by colour, not always an accurate mode of definition, the Red Men of
30 PXRUVIA SCYTHICA. America and the Swarthy Men of Australia would be inter- mediate in language to the Yellow Men of Eastern Asia and the Black Men of Africa. The Malays, or Brown Men, also present in their languages a quantity of parallels to those of the Africans, as I have noticed in my book on Numerals ; and some of these parallels affect Europe and America as well : thus the African ' threes', tal and tsala, not only re- semble the Malay ' threes', tal, tulu, and telu, but also the zal of the Etruscan dice (infra, chap, v), and the Talatui teliko in California. There are two ways of explaining African affinities like these, which crop out in parts of the globe most remote from one another. They may either be explained entirely by the primeval affinity which unites the African race to mankind in general, and to certain other races of men in particular; or they may be ex- plained, partly from this primeval affinity, and partly by the supposition, that the races who left the primitive home of mankind later than the Africans, found those Africans where they themselves settled in their migrations from the centre of the Old World, and received, in conse- quence, some African intermixture. I incline myself, while upholding the doctrine of primeval affinity and its results upon language, to believe that there was intermixture also, both in blood and speech ; a supposition on which I have endeavoured elsewhere to account for the African affinities of the Basque language : and I should conjecture that, when the Iberians migrated from Asia into Europe, they may have found there some members of an Ethiopian race allied to the modern Negroes and Kaffirs ; that the Dravidas, Chinese, Malays, and other Turanians, found in like manner a kindred race, now represented by the Melanesians, in the countries which they occupied; and that there was also, though this is more doubtful, a race of the same origin in the Arctic regions of the Old World. These supposed Arctic Ethiopians, for whom the name of ' Hyperboreans'
PERUVIA SCTTHICA. 31 might be suggested, would hardly have left any represen- tatives at the present day ; but would have been absorbed , partly by the ancestors of the Americans when they passed through North-Eastern Asia on their way to Alashka from the Central Asiatic Plateau, and partly, at a later time, if any were left, by such Turanians as spread themselves northward, like the Tungusians and the Fins. Probably none of this ancient Ethiopian race, which I have hypothe- tically divided into Africans, Melanesians, and Hyperbo- reans, would have passed into America independently, or in an unmixed state. Their connection and mixture with the Early Americans would rather have been effected in Asia itself, in the regions of Eastern Tartary and Siberia; although it is possible that some few Ethiopians may have reached the New World before the Americans, as some few Melanesians may have anticipated the Polynesians in Poly- nesia. A sculptured figure of the true Negro type is said to have been lately found in Mexico. As America and Polynesia seem, to a certain extent, parallel cases in their mode of population, I subjoin here, from De Quatrefages, the results of his inquiries relative to the first peopling of Polynesia (p. 176) : - " Les Polyn^siens se sont ^tablis et constitu^s d'abord a Samoa et a Tonga ; de la, ils sont passes dans les autres archipels de ^immense oc^an ouvert devant eux. " En abordant les lies qu'ils venaient peupler, tantot les Emigrants les ont trouv^es enti&rement ddsertes, tantdt ils y ont rencontr^ quelques rares tribus de sang plus ou moins noir, £videmment arriv^es la par quelques-uns de ces acci- dents de navigation qu'ont pu constater presque tous les voyageurs Europ^ens. " Soit purs, soit allies a ces tribus nfegres erratiques, ils ont form^ de centres secondaires d'ou. sont parties de nou- velles colonies qui ont dtendu de plus en plus l'aire Poly- n£sienne.
82 PERUVIA SCTTHICA. " Aucune de ces migrations ne remonte au deld des temps historiques. "Quelques-unes des principales onteu lieu soit pea avant, soit peu aprfes l'&re chr&ienne; d'autres sont bien plus r£centes, et il en est de tout-& fait modernes." America would have been peopled much earlier than this, as the comparison of numerals will have already shown. Five, Seven. We have now analysed and compared the Quichua nume- rals for ' one', ' two', ' three', ' four', and ' eight', as well as for 'hundred'. The next numerals which may be appro- priately considered together are ' five', and ' seven', pidka (= Aymara piska), and kandis. The -ka of pid-ka, ' five*, is most probably, at least in my opinion, no more than a suffix, like the -kay of the Quichua ys-kay, ' two', and the terminations of the lately cited ' ones' of California and Texas, pa-ka, pah-ko, and pu-ku (p. 28). Moreover, if we retrench this supposed suffix, -kay, from the Quichua ys-kay, ' two', we reduce it to ys, which is almost the same as the termination of the Quichua kand-is, seven', a numeral usually = ' five-two'. We may, consequently, with con- siderable probability, infer that kan6, as well as pid, is a Quichua five', which would make the complete form of the Quichua 'five* to be like kpaind, kpend, or quend (pronounced like the English word quench). For such a 'five' there are many parallels, of which some have been already given above (p. 8). What are now added below seem sufficient to show the affinities of the Quichua ' five': - < Five'. Quichua pid kand- Complete Quichua form kpend or quend Aryan - Pakhya pad Tharu pade
i
i
PfiRUVIA 6CYTHICA. 33 'Five'. Sanskrit pan* dan Greek pente Lithuanian penki Armenian King Latin quinque Kanyop (Senegambia) kanyan American - Chinook kwanam Unalashkan khaan Talatui kasako St. Raphael kenekus Uchee dwanhah Tesuque pahnouh Kioway onto To complete the comparison, 'five* and seven' must now bo taken in conjunction : - ' Five'. ' Seven'. Quichua pidks. kand-is, 5+2. Talatui (California) kasako kani-kuk, 5 + 2. Tesuque (New Mexico) pahnouh da-e, 5 + 2. Patagonian keyzum ku-ka, 5 + 2. Attakapa (Texas) pa-ghu, 5 + 2. Kuskutshewak (Alashka) amza-Jchvanam, 2 + 5. Unalashka (Aleutian) khaan u-kun } 2 + 5. Chinook (Oregon) kwanam Burmese Mtm-nith, 5 + 2. khwan~nh&ch, 5 + 2. Sanskrit pan* dan sap-Jaw, 5 + 2. Greek pente hep-fa, 5 + 2. Lesgi (Caucasus) ante-lgo, 5 + 2. 1 Kioway (Texas) onto pan-za, 5 + 2.
1 Ant go and ankhgo are 'six', while anUgo and anzgo are 'ten'; whence ant-, 'five', may be inferred. For 4, 'two', see p. 20.
34 PERUVIA SCYTH1CA. Here the Aryan approaches nearer to the Quichua and its American kindred than the Turanian does ; for the Aryan has the same ' five' that the Quichua has, and also employs in ' seven* the same 'two', though combining it with a different 'five', sap-, which has been explained above (p. 10), and will be noticed again when the Quichua 'six', sokta, comes nnder review. The Burmese, on the other hand, employs in ' seven' a different ' two* from the Quichua, but the same 'five': yet it has not this 'five' for the actual numeral, 'five', as the lchuwr or hhwan-, 'five', of the Bur- mese 'seven', can hardly be identified with the Burmese five, na. The Chinese has 'n, ' five', and z'at, ' seven'. The Tungusian 'five', tonga, belongs, I think, to a different class of hand-Jives from that which includes the Aryan and the Quichua ' five' (ante, p. 32) ; but it may possibly be allied to the Quichua dunka, ' ten', a numeral for which it is diffi- cult to select the right explanation and the true affinities. Ten. The most obvious parallels to the Quichua seem those given below :-- 'Ten'. Patagonian xahen Quichua dunha Kioway (Texas) cokhi Tesuque (New Mexico) taheh Tungusian guy en guan gan Turkish on Lesgi antgo Tshuvash (Russia) wonha Kurile wambi Mampa (Sierra Leone) wan Mano (Liberia) pfuh
PERUVIA SCYTHICA. 35 All these ' tens' would hardly, however, be akin to one another. The two African ' tens', for instance, which close the list, would belong to the same class of hand-five as the Quichua and Aryan 'five', and as the Bola (Senegambia) kanyen, ' five'; for these African ' tens', wan and pfun, and an African 'five' like hanyen, are to be classed with the African van, 'upper arm* (Kasm), pfen, 'leg* (Nso), and kanyen, ' arm* (Bola), which rank with the Patagonian fan, 'hand', and the Catoquina pan, 'arm', in South America. In like manner, the Quichua dunka, 'ten', is to be compared, if not classed, with the Quichua danka, ' leg*. Yet it is not easy to say if the Quichua dunka, ' ten', and danka, l leg*, supposing them to be classed together, should be referred to the same hand-five as such African words for 'five', 'ten', 'leg', 'arm', 'hand*, etc., as those just cited; and not rather to that other African hand-five which appears in tan, ' one', tan, 'five', tan, 'ten', tank, 'leg*, tanga, 'foot', donga, 'hand', gungo, ' arm', zanga, ' hand', zoana, ' five'; all which are to be considered identical with the Malay tong, tang a, ' hand'; with the Australian tona, tenna, 'foot'; with the Siamese tin, ten, 'foot'; with the Tungusian tonga, tonsa, sunga, ' five'; with the Korean sun, ' hand'; with the Kamtshatkan tono, 'hand'; and with the Ugalents zoane, 'five', near Mount St. Eiiad in North America. See ante, p. 8. With regard to the Tungusian ' tens' in the previous list, gu-yen, gu-an, and g-an, it might be suspected that their first element, gu org-, is equivalent to the Tungusian 'twos', goua, geio, and gur; and, if so, then, as 10=2 X 5, the other element in ' ten', -yen or -an, would be ' five', and probably allied to the Unalashkan khaan, 'five', etc. The Quichua d-urika,, ' ten', might possibly be explained, in the same manner, as = 2 x 5, if we regard -ka as a mere suffix, like ~kay and -ka, in ys-kay, 'two', and pid-ka, 'five': but, if we adopt this explanation, there would be no connection be- tween the Quichua words, dunka, ' ten', and danka, ' leg';
36 PERUVIA SCYTHICA. and the disregard of such a connection appears in my eyes a very important objection, as far as the Quichua is con- cerned, to the following resolution of a group of ' tens 1 : - 'Ten',=2x5. Quichua diir-nka Kioway (Texas) co-khi Tesuque (New Mexico) ta-heh Tungusian gu-yen gu-an ff-an g-a Turkish o-n Tshuvash wo-nka Kurile - Tarakai wa-mbi Yeso/a-rabe Upon the whole, I should be inclined to refer the Quichua pidka, ' five', and dunka, 'ten', to the same origin, and thus to connect them with the q-q class of hand-Jives, like the Latin quinque, the Sanskrit paAdan, and the Grison dune. One reason for such a connection is found in a Paraguayan language, the Guato, some numerals of which are allied to Quichua numerals. Thus the Guato 'two' and 'four', dououni and de-kai, may be referred, like the Quichua 'two' and ' four', to our own two and two-two. Now, in Guato, qumoida is 'ten', and bears some resemblance to the Quichua dunka, 'ten'. But, in the same Guato language, quin- enters into ' five', as well as into ' ten': for quin-oibo is ' fif-teen'; and -oibo comes out as '-teen' by comparison with 'eleven', ' twelve', ' thirteen', and ' fourteen', which are formed by adding caibo, a common South American ' hand', to ' one', 'two', 'three', and 'four'. And thus, as dekai-ai-caibo is ' fourteen', while quin-oibo is ' fifteen', quin- would be ' fif-' or 'five'. But this is not all : for quad- would also be 'five', as appears from tho following Guato numerals : -
PERUVIA SCYTHICA. 87 15. quin-oibo, 5+10. 16. denai-ai-quadoibo, 1 + 15. 1 7. dououni-ai-quadoibo, 2 + 15. 18. doum-ai-quadoibo, 8+15. 19. dekai-ai~quadoibo, 4+15. 20. quinoui-ai-quadoibo, 5 + 15. The result is, that quin- and quad- are both ' five', thus giving quaint or quend as the complete Guato ' five', just as the Quichua jn<fka, 'five', and kand-is, ' seven', gave kpench- or quend &s the complete Quichua 'five', which is thus iden- tical with the Guato. To this same hand-five is to be re- ferred the Guato quinoida, ' ten', and therefore, it may be, the Quichua dunka, 'ten*. As quin- is 'fif-' in Guato, while quinoi-da is ' ten', it is possible that quinoi-da is = 5 x 2 ; in which case we might draw the following parallels :- Guato quinoui-j r five\ quinoi-da, ' ten', 5x2. Quichua dun-ka, ' ten'. Tshuvash (Russia) won-ka, c ten'. Kurile wam-be> ( ten'. fam-be, ' ten'. Kioway (Texas) co-khi, c ten'. Tesuque (New Mexico) ta-heh, ' ten'. Tungusian g-a, ten'. ff-an, ' ten . (fu-yen, 'ten*. Patagonian xa-ken, ' ten'. It may be difficult, as I said at the outset, to decide how the Quichua dunka is to be best explained : but, setting aside the question of explanation, and taking up that of likeness only, there can be no doubt that the Quichua 'ten' nearly resembles Turanian 'tens'. That resemblance may, perhaps, be exhibited most obviously thus : -
38 PERUVIA SCYTHICA. ' Ten'. Tibetan duh du- thamba Kurile fambe wambe Tshuvash wonka Quichua du- nka Long as ' ten' has detained us, a few more observations upon it may yet be advisable. Numerals are generally de- rived from some word or words for ' finger'; and such would be the case in the New as well as in the Old World. For, in the Chibcha or Muysca language of New Granada we find the words : yti-quyn, ' finger of the hand', with yta, 'hand'; and quihid-yua, ' toe, dedo del pid', with quihida or qhiddy 'foot': from which we get both quyn and yua as terms for 'finger*. In like manner, the Camacan guan wati, ' digitus pedis', when combined with the Cotoxo lioate or uada, ' pes*, gives guan, ' finger'. ^This falls in with what I have observed in my Numerals (p. 41), that all words like the Latin quin-que, for instance, or the Uchee coon-jpah, * fingers', kean-thah, ' hand', and dwan-hah, ' five', " are ap- parently produced by the reduplication of some element like que in sound, with the frequent addition of the anuswdra termination, m or n; and that 'finger', 'member*, or 'limb', would probably b6 the original meaning of this fundamental element". Referring to the work in question for a fuller exhibition of the general evidence on this point, I may pre- sent as follows its bearing on the American languages : - Chibcha (New Granada) quihidyua, ' toe', =' foot-finger'. Bay ano (Guatemala) quendique, ' one', =' foot-finger'. Cunacuna (Darien) quensacua, ' one'. quensactta, ' one'. vocua, 'tw.o', etc. (p. 19). Chibcha (New Granada) . . -yua, 'finger' (see above).
PERUVIA SCYTHICA.
39
Tupi (Brazil) Camacan (Brazil) Chibcba (New Granada) Araucanan (Chili) Australia Africa
1 = ultimately 'hand + hand' x 'hand 4-hand': or, as 'hand'^' finger- finger 1 , and 4 finger' = 'one', i hundred ' would be an abbreviation of 4 finger', or ' one', eight times repeated.
40 PERUVIA SCYTHICA. Kioway (Texas) . . onto, ' five'. Tesuque (New Mexico) . pah nouh, 'five'. Patagonian . . fan, 'hand'. The Quichua padak, ' hundred', and dunka, ' ten', in this list, have been already compared (ante, pp. 27, 34) with the Chinese pak and peh, and the Tuschi bah, ' hundred', to which the Tibetan barky a may now be added; and with such ' tens' as the Turkish on and the Tshuvash wonka. Primeval affinity may perhaps have produced some of the following apparent analogies in various multiples of 'ten': - Quichua Awnu/millio^lOxlOx lOxlOx 10x10. Basque^, hul()xl0 eun ) Sibsagar Miri uyifi^uyinko, ' hundred', 10 X 10. uyine, 'ten'. Abor Miri uvfiko, ' ten'. Turkish on, ' ten'. Tshuvash ivonka, ' ten'. Quichua dunka, ' ten'. Kioway (Texas) cokhi, ' ten'. (Ne^SSo) } taheh > ' ten '' Chinese yik \ ' hundred thousand', yih) 10x10x10x10x10. pa * | ' hundred', 10 X 10. Tuschi (Caucasus) bah, ' hundred', 10 X 10. Tibetan barky a, 'hundred', 10x10. Gyarung (B. Tibet) par ye, ' hundred', 10 x 10. Thaksya (Nepal) bhra, 'hundred', 10x10. Quichua huaranka, l thousand*, 10 X 10 X 10. Armenian harivr, ' hundred', 10 X 10. bivr, ' myriad*, 10 X 10 X 10 X 10. Persian bayvar, ' myriad', 10x10x10x10. Zend baivare, ' myriad', 10 X 10 X 10 X 10.
PERUVIA SCYTHICA. 41 Nine. The Quichua for ' nine', yskun, seems to begin with yskay, 'two': yet I am not aware that any 'nine' has been resolved into 'two-seven'; nor is such resolution a probable one. But ' nine* can be resolved into ' four-five', as ' fourteen' is resolved into 'four-ten': and we have previously seen abun- dantly that 'four' is frequently 'two-two'. The original Value of the Quichua 'nine', yskun, may thus have been 'two- two-five', which would explain its resemblance to yshay, ' two'. And this inference is confirmed by comparing yshun with the ' nine* which comes nearest to it, the Unalashka sikhe&n. For sikheen, ' nine', is clearly resolvable into the Unalashka sikhin-khaan, 'four-five', as the Unalashka kan- kheen, ' eight', is into the Unalashka kanku-khaan, ' three- five'; this Unalashka &a?i-ku, 'three', 1+2, it may here be observed also, having given us earlier, with the aid of the kindred Esquimaux ' threes', pinga-wak and pinga-sat, the explanation of the Quichua kim-sa, 'three'; so that the Quichua and the Esquimaux have been already found to bear upon one another. Turning now from Unalashka to the other extremity of America, we see that the Patagonian keka-xezum, ' nine', is as manifestly = Patagonian kekaguy- keyzum, 'four-five', as the Unalashka sikh-cen, 'nine', is= Unalashka sikhin-khaan, ' four-five'. It is uncertain, and unimportant, whether the Quichua yshun, ' nine', should be resolved into ys-kun, ' two-five', one ' two' being deficient (cf. Unalashka u-kun, ' seven', 2+5) ; or into ys-k-un, 'two-two-five' (cf. Unalashka aUun, ' six', 1 + 5, atoken-khaan. These are the closest parallels which I have found for yshun : - ' Nine'. Papiah (Guinea) sipo Madagascar siva Timor sioh G
From the foregoing group of 'nines' we may see once more, as previously in the numerals, 'hundred', 'eight', ' four*, and ' two', how greatly Chinese numerals have ap- parently suffered from mutilation. For who could discern, without the help of other languages, and especially the American languages, that the Chinese ' nines', kiu and kau, with the Burmese 'nine', ko, may be identical with the Aryan two-two-five, duo-duo-quinque ? Yet the following comparisons show that such identity is quite possible : -
1 In resolving ys-k-un, as here, into 4 two-two-five', the vowel of the last syllable might belong, as also in the parallels beneath, to the second 4 two', and not to the ' five 1 . a For mttr, i two', see ante, pp. 19, 21. * Pehel, ' four', is found two lines above, in the Netela jptAil-enga, * nine'. 4 Observe how the Turkish ' fives' can supply the $ terminations here;
44
FXRDTIA SCYTHICA.
* Two', 'Four', < Five', < Nine*, 2. 2x2. 5. 2x2+5. Tshuvash ikke dwa-tta pilik tu-kh-ur Timbuktu ahinka ata-ki igu ya-g-a Tungusian <fur dy-gyn tunga ye-g-in dyul du-ye tonsa khu-y-u Ude pha bi-p kho WU-% Talatui oyo-ko oi-88uko kasdko oo-i Siamese son 8% hi ka-u Chinese dr sze wu Jd-u m % sze 'n ka-u Burmese nhach 1 le fia ko
The transition is easy from the Tesuque kuaenouh to the Chinese kau y nine', through the Skwali klwun, nine', and the Tungusian 'nines', uyun and khuyu, or uyen and huyu; and also from the Quichua yskun, ' nine', to the Mongolian and Tungusian 'nines', yisun and yegin, with the Unalashka sikheen in the Aleutian Islands, the stepping-stones between the Old and the New Worlds. Were it not beyond the scope of these pages, and unnecessary besides, as the work has been already done far better than I could do it, much evidence might be adduced in favour of the derivation of the Americans from Eastern Asia, in addition to that with which the analysis and comparison of numerals has now supplied us. This additional evidence may be found by any- one in Prescott, from whom I will only cite a few passages, to show how the two lines of evidence converge to the same point; Take, for instance, the following quotation which he gives from fiumboldt in reference to the physical character of the Americans : - an additional argument for the resolution of these 4 nines' into 'two-two- five'. 1 With an initial n two, like this or the Nepalese m, the Aryan 4 nine', na-v~an 9 might be reduced to f two-two-five'. But Sarmatian ' nines', like the Lithuanian de-w-yni, have d instead of n.
««
PERUVIA SCYTHICA.
45
" On ne peat se refuser d'admettre que I'esp&ce humaine n'offre pas de races plus voisines que le sont celles des Amdricains, des Mongols, des Mantchoux, et des Malais" This is quite in accordance with the following 'nines', mostly taken from the lists above, to say nothing of other numerals :-
It must, however, not be forgotten, that these linguistic coincidences extend to some African nations also, as will have been seen from the lists of 'nines'. In confirmation of what Humboldt has said of the physical character of the Americans in general, we may take what Prescott himself says of the manners, customs, and institutions of the ancient Mexicans, in the Appendix to his History : - "The probability of such a communication, especially with Eastern Asia, is much strengthened by the resemblance of sacerdotal institutions, and of some religious rites, as those of marriage, and the burial of the dead ; by the prac- tice of human sacrifices, and even of cannibalism, traces of which are discernible in the Mongol races ; and, lastly, by a conformity of social usages and manners, so striking, that the description of Montezuma's court may well pass for that of the Great Khan's, as depicted by Maundeville and Marco Polo." The following resemblance has been frequently men- tioned ; but it is so significant that I must notice it once more :- " A proof of a higher kind is found in the analogies of
46 PERUV1A SCYTHICA. science. We have seen the peculiar chronological system of the Aztecs; their method of distributing the years into cycles, and of reckoning by means of periodical series, in- stead of numbers. A similar process was used by the various Asiatic nations of the Mongol family, from India to Japan. Their cycles, indeed, consisted of sixty, instead of fifty-two years ; and for the terms of their periodical series, they employed the names of the elements, and the signs of the zodiac, of which latter the Mexicans, probably, had no knowledge. But the principle was precisely the same. "A correspondence quite as extraordinary is found be- tween the hieroglyphics used by the Aztecs for the signs of the days, and those zodiacal signs which the Eastern Asiatics employed as one of the terms of their series. The symbols in the Mongolian calendar are borrowed from animals. Four of the twelve are the same as the Aztec. Three others are as nearly the same as the different species of animals in the two hemispheres would allow. The remain- ing five refer to no creature then found in Anahuac. The resemblance went as far as it could. The similarity of these conventional symbols, among the several nations of the East, can hardly fail to carry conviction of a common origin for the system, as regards them. Why should not a similar conclusion be applied to the Aztec calendar, which, although relating to days, instead of years, was, like the Asiatic, equally appropriated to chronological uses, and to those of divination?" It is the Mexicans who are connected by these symbols and by their calendar with the Mongolian or Yellow Race. A similar link for the Peruvians may be supplied by their quvpus, " which/' says Prescott, " must be allowed to bear some resemblance to the belts of wampum - made of coloured beads strung together - in familiar use among the North American tribes, for commemorating treaties, and for other
PBRUVIA SCYTHICA. 47 purposes". 1 These knotted cords of the Peruvians per- formed for them, however imperfectly, the services of our arithmetical figures and alphabetical characters; although they were, " doubtless, a wretched substitute for the alpha- bet, . . . and far below the hieroglyphics, and even the rude picture-writing of the Aztecs/' The ancestors of the Chinese would seem to have used a similar substitute for writing before the invention of their present syllabic cha- racters : - " While the relics of hieroglyphic writing are so scarce that they barely suffice to prove that it was ever practised in China at all ; there is on the other hand a tradition re- corded by the followers of Confucius (not by Confucius himself, as is vulgarly supposed) in the classic of Changes, that knotted cords were used in the administration of government before the sages invented writing; and this tradition can hardly be called in question, seeing the same practice was found actually existing among the Peru- vians." 2 All this supplementary evidence, which has been briefly noticed in the last few pages, and which relates to physical character, manners, customs, and science, falls in remark- ably with the evidence derived from numerals, which we have now almost exhausted. It will, in fact, be completed, when we have considered the numeral ' six 9 , the last of the ten members of the decade which remains to be compared and analysed. Six. The Quichua ' six', 8oMa y has many parallels, as below ; but they cannot all be identified together : - 1 The quipus, i.e., ' knots 1 , "might be regarded as the Peruvian system of mnemonics"; and thus remind us of our own habit of tying a knot in a handkerchief. 8 Chalmers, Origin of the Chinese, p. 5.
48 PERUVIA SCTTHICA. 'Six.' Quichua sokta Tesuque (New Mexico) sih Annamitic sau Khyeng (Arraoan) sank Burmese khyauk Siamese hok Ahom (Assam) ruk Chinese Ink luh Nepalese ruk ruk-ka (suffix) ink tuk-6% (suffix) tuk-ya (suffix) 1 tu re Jehu cah Tibetan thu Circassian shw Basque sei Aryan sh sex shash achsaz kli8va8 wez whe Iroquois y&i yahyook yahiak achiak * Compare these suffixes with the supposed Quichua suffixes, -&*, kay.
PERUViA SCYTHICA. 49 ' Six'. Sioux ahkewe shahpai Houssa shida Bode (Bornu) zdu Kandin (N.W. of Bornu) shislies Berber sedis Syriac sheth Hebrew shdsh c Six* is commonly resolvable into ( five + one', or ' one+ five\ but sometimes into ( three X two', or s two X three'. If we resolve the Quichua sok-ta, ( six', into c five + one', it might be identified with the Opatoro (Honduras) saihe-ita, ' five-one*. Sok- would then be ' five', and its affinities will have been given above (pp. 10, 11). Those of -ta, 'one', will be found below : - ' One'. Quichua -ta St. Miguel (California) tohi Pujuni (California) ti Opatoro (Honduras) ita Caraho (Brazil) ita Natchez (Mississippi) witahu Uchee (Florida) sah Malay sa aisa Polynesian ta tahi Circassian zi Tuschi zha Lesgi zo Ude sa Our survey and analysis of the Quichua numerals are now finished; and it only remains to indicate the conclusions H
50
PERUVIA SCYTHICA.
which readers will probably have drawn daring the pro- gress of the examination. The Qnichna numerals, then, plainly connect the ancient Peruvians with the nations of the Old World; their nearest kindred there appearing to be the Turanians, especially those of the Yellow Race. But the connection of the Peruvians with the Americans in general is still closer than that with the Turanians : and, if it were required to point out any particular American nations as coming nearest to the Quichua in the largest number of numerals, the Tesuque of New Mexico, and the Kioway of New Mexico and Texas, might be selected from the rest, although the identity in numerals may frequently be hardly perceived without recurring to the previous analysis. I conclude, however, this chapter on numerals by placing the three decades side by side, and add to them the Patagonian decade : -
South America. Nobth America.
Quichua. Patagonian. Kioway. Tesuque.
i. huh cheuquen pahko guih
ii. yakay xeuk&y 1 gia kihyeh
in. kvmaa keash pao pohyeh 1+2. IV. ttahua kekaguy iaki ionouh 2x2. v. pi<fk& keyzum onto pahnouh
vi. sokta wenecash 3 mosso sih
vti. kandis kuka panza dae 5+2. VIII. pU88ak wenekekagne iaza kuhbeh 2x2x2 ix. yskun kekaxezum cohzu kuaenouh 2x2+5 x. dunka xaken cokhi talieh
1 In these Patagonian numerals, x is the Spanish x, and therefore equivalent to a German ch. * As already noticed, wenecash, 'six', =2x3, as wenekekagne, * eight'.