GN 480 .5 .L6 Copy 1 UNIVERSITY OF CALIFORNIA PUBLICATIONS ~ IN AMERICAN ARCHAEOLOGY AND ETHNOLOGY Vol. 16, No. 2, pp. 29-45 March 29, 1919 THE MATRILINEAL COMPLEX BY ROBERT H. LOWIE UNIVERSITY OF CALIFORNIA PRESS BERKELEY Sfono^apn UNIVERSITY OF CALIFORNIA PUBLICATIONS DEPARTMENT OF ANTHROPOLOGY The following publications dealing with archaeological and ethnological subjects issued under the direction of the Department of Anthropology are sent in exchange for the publi- catlona of anthropological departments and museums, and for journals devoted to general anthropology or to archaeology and ethnology. They are for sale at the prices stated. Exchanges should be directed to The Exchange Department, University Library, Berkeley, California, U. S. A. AH orders and remittances should be addressed to the University of California Press. AMERICAN ARCHAEOLOGY AND ETHNOLOGY.— A. L. Kroeber, Editor. Prices, Volume 1, $4.26; Volumes 2 to 11, inclusive, $3.50 each; Volume 12 and following, $5.00 each. .-Cited as Univ. Calif. Publ. Am. Arch. Ethn. Price Vol. 1. 1. Life and Culture of the Hupa, by PUny Earle Goddard. Pp. 1-88; plates 1-30. September, 1903 $1.25 2. Hupa Texts, by Pliny Earle Goddard. Pp. 89-368. March, 1904 3.00 Index, pp. 369-378. Vol. 2. 1. The Exploration of the Potter Creek Cave, by William J. Sinclair. Pp. 1-27; plates 1-14. April, 1904 ; 40 2. The Languages of the Coast of California South of San Francisco, by A. L. Kroeber. Pp. 29-80, with a map. June, 1904 60 3. Types of Indian Culture ta California, by A. L. Kroeber. Pp. 81-103. June, 1904 „ 25 i. Basket Designs of the Indians of Northwestern California, by A. L. Kroeber. Pp. 105-164; plates 15-21. January, 1905 75 5, The Yokuts Language of South Central California, by A, L. Kroeber. Pp. 165-377. January, 1907 2.25 Index, pp. 379-392. Vol. 3. The Morphology of the Hupa Language, by Pliny Earle Goddard. 344 pp. June, 1905 3.50 Vol. 4. 1. The Earliest Historical Relations between Mexico and Japan, from original documents preserved in Spain and Japan, by Zelia Nuttall. Pp. 1-47. April, 1906 50 2. Contribution to the Physical Anthropology of CaUfomia, based on collec- tions in the Department of Anthropology of the University of CaUfomia, and in the U. S. National Museum, by Ales HrdUcka. Pp. 49-64, with 5 tables; plates 1-10, and map. June, 1906 75 3. The Shoshonean Dialects of California, by A. L. Kroeber. Pp. 65-166. February, 1907 1.50 4. Indian Myths from South Central California, by A. L. Kroeber. Pp. 167- 250. May, 1907 _ 75 5. The Washo Language of East Central California and Nevada, by A. L. Kroeber. Pp. 251-318. September, 1907 75 6. The Religion of the Indians of California, by A. L. Kroeber. Pp. 319-356. September, 1907 50 Index, pp. 357-374. VoL5. 1. The Phonology of the Hupa Language; Part I, The Individual Sounds, by Pliny Earle Goddard. I^. 1-20, plates 1-8. March, 1907 35 2. Navaho Myths, Prayers and Songs, with Texts and Translations, by Wash- ington Matthews, edited by Pliny Earle Goddard. Pp. 21-63. Septem- ber, 1907 75 3. Kato Texts, by Pliny Earle Goddard. Pp. 65-238, plate 9. December, 1909 2.50 4. The Material Culture of the Klamath Lake and Modoc Indians of North- eastern California and Southern Oregon, by S. A. Barrett. Pp. 239-292, plates 10-25.^^une, 1910 75 5. The Chimariko Indians and Language, by Roland B. Dixon. Pp. 293-380. August, 1910 1.00 Index, pp. 881-384. Vol. 6. 1. The Ethno-Geography of the Porno and Neighboring Indians, by Samuel Alfred Barrett. Pp. 1-332, maps 1-2. February, 1908 _ 3.25 2. The Geography and Dialects of the -Miwok Indians, by Samuel Alfred Barrett. Pp. 833-368, map 3, 3. On the Evidence of the Occupation of Certain Regions by the Miwok Indians, by A. L. Kroeber. Pp. 369-380. Nos. 2 and 3 in one cover. February, 1908 50 Index, pp. 381-400. ^H' «■ L.*- UNIVERSITY OF CALIFORNIA PUBLICATIONS ^ AMERICAN ARCHAEOLOGY AND ETHNOLOGY <^ Vol. 16, No. 2, pp. 29-45 March 29, 191S THE MATRILINEAL COMPLEX EGBERT H. LOWIE Dr. Hartland has recently reopened discussion of a question which has for a number of years been regarded as closed by American ethnologists — the relative priority of matrilineal and patrilineal insti- tutions.^ It is always desirable to reexamine from time to time the fundamental conceptions of a science, and a challenge to accepted theories often leads to a remarkable illumination of basic principles. The main objection to Dr. Hartland 's essay on "Matrilineal Kinship and the Question of Its Priority" must rest not on his refusal to bow to the established American view, but on his inadmissible method of approach. Dr. Hartland defends two important propositions. In the first place, he believes that "normally and apart from a few exceptions that seem well established, kinship was originally reckoned on one side only" (p. 24). Secondly, he contends that descent through the mother regularly preceded descent through the father. The first of these contentions seems to me singularly ill-founded, inasmuch as we find almost uniformly that the tribes on the lowest level of civilization, whether Andaman Islanders, Sakai, or Plateau Shoshoneans, lack the unilateral mode of reckoning kinship. However, at present I am con- cerned solely with the second of Dr. Hartland 's propositions. But before entering into a discussion of his method of proof, I must deal with a matter of terminology. In America it has been customary of late to refer to matrilineal social units as "elans" and to patrilineal groups as ' ' gentes. ' ' This involves the unfortunate lack of a generic term for a unilateral group regardless of mode of descent. Moreover, such usage conflicts both with Lewis H. ]\Iorgan's use of gens in the generic sense, and the generic use of clan firmly established 1 E. S. Hartland, Matrilineal Kinship and the Question of its Priority, Mem. Am. Anthr. Assoc, iv, 1-90, 1917. 30 Unii-crsitt/ of California Publications in Am. Arch, and Ethn. [Vol. 16 among English writers. In an elementary course on anthropology at the University of California, I eliminated these difficulties by con- sistently employing the term "kin" generically. and "mother-kin" or "father-kin" by way of specification. Since then, however. Drs. Goddard and Kroeber have i)ointed out the misleading connotations of tlie term kin when tcclmically restricted to the unihiteral (and normally exogamous) group. Accordingly. I will substitute the old term "sib," which has recently been resuscitated in Professor Phil- brick's translation of Huebner's History of Germanic Private Law. How, then, does Dr. Hartland establish the conclusion that existing father-sibs have grown out of mother-sibs, thus converting an observed simultaneity into a chronological sequence ? His own statements leave no doubt whatsoever as to his method of procedure. He determines first "what are the chief cliaracteristics of the matrilineal organization of society" (p. 7). This is accomplished "by taking a people in which that organization is exhibited in tlie full strength and ni)ting its peculiarities" (ihid.). When subsequently such features are en- countered in combination with patrilineal descent they are interpreted as "survivals of matrilineal polity" (p. 23). The logical error involved in this procedure is patent. Dr. Hart- land is obliged to introduce in the place of mere matrilineal descent, about which the discussion revolves, the very different concept of a matrilineal complex; and that complex he establishes not by empirical observation but by selecting a people in which it is supposed to be exhibited in its full strength. This estimate as to the vigor of matri- lineal organization is clearly arbitrary; Dr. Hartland has rational- istically constructed an organization such as might logically follow from matrilineal descent and then finds a few concrete illustrations of this purely a priori conception, from which in turn he deduces the traits of the mother-sib. The task of the critical ethnologist is very different. Starting from the one pivotal feature of maternal descent, he must establish by empirical observation what other features appear in combination with the mother-sib. This is the only pos-sibility of establishing the facts in the case. Now what are the traits which Dr. Hartland deduces as symp- tomatic t)f the typical matrilineal organizations? Essentially his enumeration (p. 10) coincides with Tylor's earlier statement.- According to both writers, the mother-sil), defined by matrilineal descent, is further distinguished by matrilocal residence; the inherit- 2 Jour. Enthr. Inst., xvni, 252, 1880. ^^'^.^ ' .^r■^- ' y^^ ' J■SsJ^ 1919] Loirie: The ilatrilineal Complex 31 ance of property within the sib ; and matrilineal authority vested more particularly in the mother's brother. In the present paper I shall discuss first the alleged correlation between matrilineal descent and matrilocal residence, and shall supplement this inquiry with a corre- sponding examination of the avunculate and matrilineal inheritance, two institutions which are best considered in conjunction. Starting our survey with North America, we find four regions with matrilineal descent — an appreciable part of the Atlantic population (embracing notably the Iroquois and the Southeastern tribes) ; three Northern Plains tribes, the Mandan, Hidatsa, and Crow; the Pueblo Indians; and some of the Northwest Coast Indians. What are the ascertainable facts with reference to residence? As to the Iroquois, Morgan's statements are hardly sufficiently explicit, but they suggest that the bride took up her abode with the groom's relatives. Our author represents the bride as conducted to the home of her intended husband, where she presents some bread to her mother-in-law as proof of her domestic accomplishments, while the husband's mother returns some venison to the girl's mother "as an earnest of his ability to provide for his household. ' '^ On the other hand, the Southeastern tribes obviously practiced matrilocal residence to .some degree. Dr. John R. Swanton informs me that among the Creek the women stayed in one place and their husbands came there from other localities, the houses of women of the same clan being built in immediate proximity to one another. This scheme, according to the same authority, seems to have prevailed likewise among the Timucua of Florida. Similarly, the Choctaw men of Bayou Lacomb, Louisiana, lived in their wives' villages.* Among the Yuchi there was no obligatory rule. A woman normally left her home and the husband built a house for the new couple ; but ' ' sometimes the man goes to live with his wife's parents xmtil he is able to start for himself."'^ For the three Plains tribes the data are fairly definite. The IMandan youth often remained in his father-in-law's lodge, but fre- quently a new hut was constructed." Similarly the Hidatsa had no absolute rule, though in the hegimting the young couple generally remained with the wife's parents, the husband acting as their servant 3L. H. Morgan, League of the Ho-de-no-saii-nee, Lloyd ed., i, 313, 1904. 4 D. I. Bushnell, The Choctaw of Bayou Lacomb, St. Tammany Parish, Louis- iana, Bur. Am. Ethn., Bull. 48, p. 27, 1909. 5 F. G. Speck, Ethnology of the Yuchi Indians, p. 95, 1909. Italics inserted by the author. 6 Maximilian, Eeise in das innere Nord- America in den Jahren 1832 bis 1834, II, 128, 1839. 32 University of California PiihJications in Am. Arch, and Etliii. [Vol. 16 and hunter/ Among the Crow, on the other hand, wedkiek generally began with patrilocal residence.* The Pueblo Indians of Arizona and New Mexico form the classical example of matrilocal abode. Among the Hopi the house belongs to the woman, and the daughter after marriage lives with her husband under her mother's roof.® The identical scheme prevails among the Zufii^" and the Sia." But this is not the usage of the nomadic Navaho : "In the absence of the husband," say our most trustworthy authori- ties, "the mother pays her daughter an occasional visit."'- The Apache custom differs from this, yet without conform-ing to the Pueblo practice. "The young man lived with his father-in-law for sowe time and hunted for the support of the family."'^ There remain the Pacific Coast people. According to Krause. the Tlingit had both matrilocal and patrilocal unions, while Swanton's account strongly suggests the preponderance of the latter.'* The Haida data are unusually illuminating. A boy became engaged be- tween fifteen and eighteen and during the period of hetrothal he lived with his fiancee's family, working for them until his marriage. But at the wedding ceremony the father of the girl politely disparaged her abilities, adding that "he knew that her future mother-in-law would take care of her, he was glad that his daughter was going to live with the young man's sisters," etc'^ y^^ the Tsimshian we have recent information to the effect that "the In-ide is carried down to the canoe. and she departs with her husband to his village, where they live. If the groom belongs to the same village, the couple often stay with the girl's parents."'*^ The facts for North America are readily summarized. :\latrilocal residence in an unequivocal form exists only in two matronymic centers— among the Pueblo Indians and among the Creek. Elsewhere such practice is confined either to the earliest period of wedlock or the preceding condition of betrothal, and bears on its face the clearest - Lowie, Notes on the Social Organization and Customs of the Mandan, Hidatsa, and Crow Indians, p. 46, 1917. 8 Lowie, Social Life of the Crow Indians, p. 223, 1912. 9 W. Hough, The Hopi Indians, p. 127, 1915. 10 M. C. Stevenson, The Zufii Indians, p. 30.5, 1904. 11 M. C. Stevenson, The Sia, p. 22, 1894. 12 The Franciscan Fathers, An Ethnologic Dictionary of the Xavaho Lan- o-uage, p. 449, 1910. 13 p E Goddard, Indians of the Southwest, p. 162, 1913. 14 A Krause, Die Tlinkit-Indianer, p. 220, 1885; Swanton, 2fith Ann. Rep. Bur. Am. Ethn.,'p. 428, 1908. ^ 15 J. R. Swanton, Contributions to the Ethnology of the Haida, pp. oO, 51, 1905. 16 F. Boas, Tsimshian Mythology, p. 532, 1916. 1919] Loicic: The MatrUiiieal Complex 33 evidence of association with a rendering of services by way of com- pensation. IMoreover, patrilocal residence occurs among tribes with mother-sibs, and in a number of instances both modes of residence exist side by side without any suggestion that either is deemed preferable. It is very interesting to note that while mother-sibs are not in- frequently consistent with patrilocal residence, a patrilineal scheme or loose organization often appears with matriloeal residence or indi- cations thereof. Thus the Blaekfoot felt that the father-in-law was for a time entitled to part of the spoils of the chase and war, especially the latter.i' Similarly, without actual matriloeal residence, the Omaha husband labors on behalf of his father-in-law for the period of one or two years.i' With the Arapaho the new couple occupy indeed a tent of their own, but it is pitched by the lodge of the bride's father.^^ Cheyenne usage seems to be strictly parallel to that of the Arapaho.-"' Among the Gros Ventre a bridegroom often settled with his father- in-law." The Eastern Dakota practiced both customs with apparently equal frequency; and the same applies to the Assiniboine." Matri- loeal residence as a normal usage of the Eastern Cree is vouched for by some of the early travelers.^^ Finally may be cited some— though by no means all— of the Central Eskimo communities, where house- keeping regularly begins with the bride's family." This list, which could undoubtedly be materially increased, will of course be greeted by adherents of the good old school as so many symptoms of a former mother-sib scheme. An auxiliary hypothesis can always be framed to account for disconcerting facts. We, how- ever, are concerned here with ascertaining the empirical data without encumbering our statement with any questionable assumption; and accordingly, our survey establishes the indisputable fact that many matrilineal tribes practice patrilocal residence, while on the other hand, some form of matriloeal residence is frequently linked with father-sibs. 17 C. Wissler, The Social Life of the Blaekfoot Indians, p. 10, 1911. IS A. C. Fletcher and F. La Flesehe, The Omaha Tribe, p. 324, 1911. 19 A. L. Kroeber, The Arapaho, p. 12, 1902. 20 E. S. Curtis, The North American Indian, vi, 157, 1911. 21 Kroeber, Ethnology of the Gros Ventre, p. 180, 1908. 22 S. E. Biggs, Dakota Grammar, Texts, and Ethnology, p. 20.5, 1893; Lowie, The Assiniboine, pp. 40, 41, 1909. 23 A. Skinner, Notes on the Eastern Cree and Northern Saulteaux, p. 37, 1911; Franklin, Journey to the Shores of the Polar Sea, m, Everyman's Library ed., p. 66. 24 Boas, The Central Eskimo, p. 579, 1888. 34 University of California Publications in Am. Arch, and Ktlin. | \'ol. IG The conclusions reached for North America must be tested by data from other divisions of tlie j:!;lobe. Turning next to I\Ielanesia as one of tlie best studied regions of Oceania, we find that in the groups investigated by Codrington tlie young man regularly takes his wife to his own or to his father's house." To Dr. Rivers' Oceanian researches we are indebted for a quite general statement on the subject : . . . there is little doulit it is usual throughout Melanesia for a iiiarrieil eoujile to live with the husband's peoide. . . . There is thus evidence that even in the ]iart of Melanesia whicdi has social institutions of tlie most archaic kind, there is no association of niatrilocal marriage v^-ith niatrilineal descent. 26 What is true for Melanesia holds so generally in Australia that Dr. Hartland is constrained to admit "the practically universal cus- tom of taking the wife to reside with her husband."-' Finally, we may consider the data from Africa. Unfin-tunately this still remains for sociological purposes the Dark Continent. At least I have failed to gain a comprehensive picture of rules of descent and residence and am obliged to present random findings. Tn that portion of the Ewe nation visited by Ellis matronymy was coupled with patrilocal abode.-* Of the Bantu the Bakongo are likewise matronymic and patrilocal.-" and this applies also to the Herero.^** The Ovambo differ from their neighbors inasmuch as female descent is here associated with a preliminary matrilocal residence during which the wife's parents are masters of the situation; liut when the young husband is alxjut t])irty he establishes a settlement of his own and gains his independence. ■■' Finally, I may cite the Makonde case from East Africa, where a young man marries his maternal uncle's daughter and lives near her father. '^ We may now summarize our total results. The Australian and Melanesian facts lend no support whatsoever to the theory that maternal descent is regularly accompanied by the matrilocal factor. The African and American data are slightly more favorable but by no means warrant the dictum that matrilocal residence is a symptom of matronymy. 25 E. H. Codrington, The Melanesians, pp. 238ff., 1891. 26 W. H. E. Elvers, The History of Melanesian Society, n. 126, 1914. 2" Hartland, op. cit., p. 65. 28 A. B. Ellis, The Ewe-speaking Peoples of the Slave Coast of West Africa, pp. 157, 207, 1890. 29 A. Weeks, Among the Primitive Bakongo, pp. 145-147, 1914. sow. Schinz, Deutseh-Siidwest-Afrika, pp. 163, 172, 1891. silbid., pp. 304, 311. 32 K. Weule, Negerleben in Ostafrika, p. 383, 1909. ]()19] Loirie: The MatrUiiical CompUr 35 This t'onelusion does uot oblige us to abandon altogether Trior's suggestion that mode of residence is connected with rules of deseent.^^ Possibly in a number of instances the retention of a girl by her family after the Ilopi and Ziuli fashion led to reckoning her children as members of the group of the house owner. But since that retention is so rarely protracted beyond the initial stage of wedlock, the hypothesis, with all its seductive plausibility, seems to be of limited applicability. Perhaps it would be better to divide sharply cases of permanent and of temporary abode with the woman's kindred. We might then find that the former category is uniformly, or nearly so, associated with matronymy. But in what part of the world except the southwest of North America and possibly the Khasi of Assam does permanent matrilocal residence occur? At present it therefore seems best to lump together all our cases under a single heading and make some estimate of the strength of the tested correlation. There is so much difficulty in weighting our geographical units and the distinction between temporary and permanent matrilocal residence that I will refrain from venturing on a mathematical computation. But as a guess I should say that the coefficient, instead of approximating one hundred per cent would be much nearer to ten per cent on the most favorable view of the case. Let us next turn to the customs embraced luider the term " a^oinculate. " In what sense is it possible to treat these as symp- tomatic of the matrilineal complex? That is, to what extent are mother-sibs connected with avuncular authority or an altogether dis- tinctive relationship between mother's brother and sister's son? The avunculate in North America is described by INIorgan in a significant passage : He is, practically, rather more the head of his sister's family than his sister's husband. . . . Amongst the Choctas, for example, if a boy is to be placed at school his uncle, instead of his father, takes him to the mission and makes the arrangement. An uncle, among the Winnebagoes, may require services of a nephew, or administer correction, which his own father would neither ask nor attempt. In like manner with the lowas and Otoes, an uncle may appro- priate to his own use his nephew's horse or his gun, or other personal property, without being questioned, which his own father would have no recognized right to do. But over his nieces this same authority is more significant, from his participation in their marriage contracts, which, in many Indian nations, are founded upon a consideration in the nature of presents. s* 33 .Tour. Anthr. Inst., xviii, 258, 1880. 3+L. H. Morgan, Systems of Consanguinity and Affinity, p. 158, 1871. 36 University of California Publications in Am. Arch, and Ethu. [Vol. 16 With reference to the AVinnebago, ^lorgan's statement has since been verified and supplemented. A man can take liberties with his maternal uncle which are exjjressly prohibited with his paternal uncle and aunt and his maternal aunt. Yet in spite of this freedom a man and his maternal uncle stand in particularly close relationship, the former always acting? in the capacity of a servant. On the war-path, particularly, this relationship is shown in its strongest jihase, for then the nephew . . . must accompany him as a sort of esquire and suffer himself to be slain should his maternal uncle ... be slain or eaptured.^s According to a remark of Tom Bear to the present writer, the Winnebago nephew may appropriate any part of his uncle's property. For another Siouan tribe we likewise possess corroborative data. Among the Omaha the nephew was permitted to jest familiarly with his uncle; on the other hand the maternal uncle had full control of the children after the parents' death and even during their lifetime was "as alert as their father to defend the children or to avenge a wrong done them.''^" Unpublished data by Murie indicate like usages among the neighboring Pawnee, while Skinner's observations indicate that the Menomini have a usage somewhat similar to the Winnebago.^'' Now it should be noted that of all the examples of the avunculate cited above, the only one to the point is that of the Choctaw. All the other tribes mentioned are either patrilineally organized or, as in the Pawnee case, lack a definite sib system. If, instead of looking for evidences of peculiar avuncular relations, Ave correlate mother-sibs and the avunculate we get the following results. Among the Iroquois there is no evidence of the avunculate, while for the southeastern peoples we have Morgan's statement as to the Choctaw. In the Northern Plains group traces of the custom are lacking. Southwestern tribes vary in their practices. In the Ilopi household the mother's brother certainly plays an important role, especially with reference to ceremonial matters; and this remark applies equally to the Tewa enclave in Hopiland.^* On the other hand, I can find no indications that corresponding customs are shared by the Navaho or even the Zuiii. The one perfect illustration of the avunculate in connection with maternal descent in North America is furnished bv the North 35 P. Eadin, Am. Anthr., n.s. xii, 213, 214, 1910. 36 J. O. Dorsey, 3d Ann. Rep. Bur. Am. Ethn., pp. 265, 270, 1884; Fletcher and La riesche, ibid., 27th Ann. Rep., p. 323, 1909. 37 Skinner, Social Life and Ceremonial Bundles of the Menomini Indians, p. 20, 1913. 38 The writer's observations; B. W. Freire-Marreeo, Am. Anthr., n.s. xv., 281, 282, 1914. 1919], Lowie: The MatriUncal Complex 37 West Coast tribes, where the nephew lives with his uuele, works for him, marries his daughter (or, it may be, his widow) and is regarded as his successor.^'' Some tendency for avuncular customs to appear with matronymy is thus apparent, but in other cases they are lacking; and they even appear with father-sibs. Since we are interested in an empirical determination of the facts, the popular theory of survivals as to the last mentioned group of cases is inapplicable as it was in our parallel findings with reference to residence. Turning from America to Melanesia, we have abundant evidence of the avunculate among those natives of this region who have been most thoroughly studied. For example, we find that in the Banks Islands the nephew obeys his maternal imcle more readily than his father and treats him altogether with greater reverence; at one time he was, indeed, the legitimate heir of his possessions and was even entitled to appropriate whatever he desired of such property during his uncle's lifetime. Similar customs are noted in the New Hebrides and Torres Islands, but it is not a little remarkable that the highest development of relevant usages is represented by the vasu institutions of the non-matronymic Fijians.*" "When we discover a hardly less pronounced avunculate among the Polynesian Tongans and distinct traces of the custom among the Saraoans, our knowledge of tribal relations suggests an interpretation very different from that of current survivalist dogmatism. If Fiji forms one center of diffusion for the practice, then its relative strength in Tonga and Samoa is precisely what we should expect on the theory of borrowing. In short, the Tongans and Samoans display avuncular features not because they ever passed through a matrilinear stage but because they have been in contact with a people where the avunculate flourished to an exces- sive degree. It would. I think, be rash to deny categorically that in certain parts of jMelanesia where mother-sibs are not observable avuncular practices are survivals of a one-time matrilineal system. This may even apply to Fiji, though this seems more problematical. However, it is worth while to contrast survivalist logic as applied to the ]\Ielanesian and the North American field. In Melanesia we find definitely matrilineal peoples practicing avuncular customs in logical consonance with their social organization. Consequently, when other members of the same 33 Boas, 31st Ann. Eep. Bur. Am. Ethn., p. 42-5, 1916. ■io Elvers, The History of Melanesian Society, i, 37, 204, 160. 38 Vniversity of California Publications in Atn. Arch, and Ethn. [Vol. 16 linguistic and cultural group show these customs without the fre- quently associated type of organization it is not improbable that that type once existed where it is no longer observed. But the North American case differs foto coeJo from this. In the Siouan family, for example, it is precisely the itiatrilineal groui)s that lack, and the patrilineal ones that exhibit, avuncular features; and the same applies, if we sliift the comparison, from the Siouan stock to the Plams culture area. Tlie logic of the two cases is thus very different. Altogether I may register my opinion that ^lelanesia is the one part of the globe w^here the substitution of a patrilineal for a matrilineal system has been fairly well established. In theoretical discussions of social organization, data from Aus- tralia play a ludicrously disproportionate part. No doubt the psycho- logical effect wrought by a thick as compared with a moderately sized volume and by books issued from the press of commercial publishers as contrasted with the. monographs due to scientific institutions is largely to blame. However tliis may be. it cannot be too vehemently or too often stated that our knowledge of the i.sland continent is extremely inadequate. Spencer and Gillen give us satisfactory infor- mation on two or at most three tribes; while Ilowitt's work is for the most part a i>ioneer's compilation, commendable a.s a first skimming of the ground, but hardly more. To be sure. Roth's studies on Queensland are excellent and A. R. Brown's researches in West Australia give promise of what trained inquirers will ultimately achieve. But altogether Australia is remarkably little known and the theorizer would do well to wait for the field worker's garnering of facts. Accordingly, it is not possible to give a comprehensive view of the maternal uncle's place in Australian society. That in various communities definite social functions belong to him. is certainly true; but these are not limited by any means to matrilineal groups, and on the other hand similar functions go with entirely different relation- ships. I find no trace of matrilineal inheritance or succession to office, and the only suggestion of avuncular authority reported from matronymic groups Mes in the right of betrothal exercised by the mother's brother over his niece among the Dieri and two or three even less known tribes.*^ Of the altogether unique avuncular relationship recorded, for example, among the Tlingit. the Banks Islanders, and the Thonga, nothing seems to be known in Australia. 41 X. A\'. Tlioiii;is, Kinship Oi-fjanizations and Gi-ou|i ^farriage in Australia, p. 22, ]!)06. 1919] I.oirie: The MalriUiicul Complex 39 Filially we may turn to Africa. Avuncular institutions have been recorded from various parts of this continent and doubtless from many tribes besides those for which I have found definite data. Of the Southern Bantu the Herero have matrilineal inheritance in such a form that while the brother of the deceased is the first claimant, the sister's son becomes heir in the absence of brothers.*- For two of the eastern tribes we have very specific data. With the Yao inheritance is from uncle to sister's son, while among the Makonde we find in addition that the mother's brother must grant his consent to a girl's marriage and is entitled to' a portion of the bride price. *^ The Bakongo regard a woman's eldest brother as master of her children, while the nephews succeed to the uncle's property and, brothers failing, to his office.** In Upper Guinea the Anglo-Ewe grant greater prerogatives as to children to the maternal uncle than to the father.*^ Since the nephew is the heir apparent his uncle expects in return adequate work during his lifetime. The boy must accordingly accompany the uncle on his travels, carrying provisions, cowrie shells and objects for barter. Incidentally he acquires the art of trading, the technique of weaving, and other useful accomplishments. Here we are again confronted, however, with the fact that insti- tutions identical or very similar flourish in equal measure among sibless or patronymic groups. Thus, the relations between mother's brother and sister's son are peculiarly intimate in the Hottentot country — closer than any except those obtaining between parents and children. To be sure, there has not been observed any matrilineal inheritance rule, but the uncle is at liberty to appropriate any of his nephew 's damaged property, while the sister 's son indemnifies himself by freely seizing perfectly uninjured possessions of his uncle. For example, while a man had taken his nephew's horse, which had defective hoofs, the young man coolly appropriated by way of com- pensation a milch cow, her calf, and ten goats.*" The altogether unique position of the malum e in Thonga society has become familiar through Junod's fascinating account. Here the mother's brother lays claim to a portion of the bride price and plays an important part in ceremonial activities, while the nephews exercise t"as«-like privileges, *2 Sehinz, op. cit., p. 178. •*3 Weule, Wissenschaftliehe Ergehnisse nieiner ethnographischen Forschungs- reise in den Sudosten Deutsch-Ostafrikas, pp. -58, 96, 97, 124, 1908. -** Weeks, op. cit., p. 107. 45 G. Hartter, Sitten und Gebrauehe der Angloer, Zeitsehr. f. Ethn., xxxvm, 4.3, 1906. 4«L. Schultze, Aus Namaland und Kalahari, p. 303, 1907. r 40 University of California Publications in Am. Arch, and Ethn. [Vol. 16 being permitted to appropriate his food, and may even inherit one of his wives.*' With the presumably Ilamitic Nandi the maternal uncle must give his consent before the boy is svibjected to circumcision or other bodily mutilations ; he normally receives a cow when his nephew has imdertaken a successful raid ; and his curse is believed to have the most deadly effect. "The most terrible thing that can happen to a Nandi is to displease his maternal uncle. ' '*^ Summing up the facts relating to the avuneulate. we are again driven to the conclusion that a ten per cent correlation probably is all that can be demonstrated on empirical grounds. For an empirical proof of Dr. Hartland 's contentions, the avuncular institutions are too frequently lacking in matronymic communities, they are far too fre- quently combined with a patrilineal scheme. It is only by assuming beforehand the theory that is to be proved, that the latter group of data can be construed into tlie opposite of their face value. As a matter of fact, no less than three distinct alternatives to the survival theory suggest themselves with reference to the avuncular customs when imbedded in a patrilineal complex. In the first place, instead of pointing to a pristine matrilineal society they may merely represent borrowed elements dissociated from the particular matri- luaeal context that occurs in a neighboring group. I have already illustrated this assumption with IMelanesian data. An additional example is furnished by a Papuan tribe. The Kai are not divided into exogamous sibs of any kind, yet a keen missionary observer notes that maternal uncles are entitled to the bride price and continue to exercise control over their niece ; that, moreover, while siTCcession to chieftainship is from father to son, the sister's son takes office when issue is lacking.*^ IMatronymic tribes possessing the characteristic features found occur in such close proximity to the Kai that trans- mission readily accounts for the phenomena. The second alternative has been suggested by Dr. Rivers. "Where the avuneulate is linked with cross-cousin marriage of the more common type, the question arises whether the altogether peculiar relationship between uncle and nephew does not simply result from that between a man and his prospective son-in-law. It is not difficult to understand that a very special bond would unite a boy with the father of his future wife. This explanation is naturally of restricted application but merges into an interpretation of generalized type. 47 H. A. Junod, Life of a South African Tribe, i, 44, 226, 253, 212, 255, 2( 1912. 48 A. C. Hollis, Tlie Nandi, p. 94, 1909. 40 C. Keysser in Neuhauss, R., Deutseh-Neu-Guinea, iii, 85-89, 100, 1911. 1919] Lou-ie: The MatrUineal Complex 41 In almost every case where primitive tribes have been exhaustively studied it has been found that various relationships are associated with definite rights and duties. The avuncular relationship forms only 02ie of a whole series of more or less analogous relationships and must be so viewed lest its importance be grossly exaggerated. Thus we find that among the "typically" matrilineal Hopi the naming of a child is a prerogative of the father's female relatives. Is this a survival of some earlier patrilineal society subsequently superseded by present conditions through some Amazonian coup d'etat f We might easily contribute to the stock of anthropological romance by develop- ing this hypothesis in some detail and should only be employing the type of logic popular among advocates of matrilineal priority. Or are we perchance face to face with a transitional condition through which the Hopi are beginning to grope towards father-right? No assumption could be less founded in reason. The Hopi are as matronymic and avuncular as they ever were; and the chances are that the naming custom is as old as any of these other institutions. Similar facts may be cited for the Hidatsa and Crow. With these strictly matronymic peoples the paternal relatives nevertheless play a perfectly definite part in the individual's social life. A Crow treated his father's brothers and other clansmen with respect and regularly invited them to feasts. When an occasion arose for giving away presents, the father's brothers and sisters were considered as recipients before every one else. When a man returned from a successful raid, he gave away some of his horses to a father's clansman. The sons and daughters of a father's clansman were the joking-relatives pos- sessed of altogether distinctive privileges. Nicknames were derived from the actions of a father's clansman, and so might be also names of honor. The father 's clansmen rejoiced over a young man 's success in war and would chant laudatory songs. Among the closely related Hidatsa the number of patrilineal functions is even greater. In addition to practically all the above mentioned usages we find the father's clan -mates conducting the funeral and bestowing new names; while in the series of graded societies individuals purchased regalia from a father's clan-mate. The emphasis on the father's side of the family is so strong among these two tribes that it might plausibly be exploited on behalf of the hypothesis that the Crow and Hidatsa were once organized into father-sibs. The real explanation is, of course, quite different. In his discus- sion with McLennan, Lewis H. Morgan pointed out the misleading implications of the phrase "kinship through females only." Every 42 University of California I'ubUralioiis in Am. Arch, and Etliti. \ Vol. 16 tribe, lie sliowed, regardless of the mode of deseeiit, reeog'iiize.s kinship in both lines, as their nomenclature clearly demonstrates. Now this purely terminological contention of ]\Iorgan's admits of wider applica- tion. Not only do people uniformly recognize the existence of bilateral relationsliips by an appropriate nomenclature, but they further assign definite duties and privileges to both sides of the family. Thus, it happens that the matrilineal Crow show a peculiar regard for the father's clan-mates, while various patroiiyinie tribes assign peculiar functions to the mother's brother. A more thoroughgoing investiga- tion in the field will reveal innumerable social functions dependent on a special type of relationship, patrilineal or matrilineal, by blood or by marriage. The avunculate cannot be appraised rightly except as a special case of a very general tendency to associate definite social relations with definite forms of kinship regardless of maternal or paternal side. The explanation here offei-ed may be supplemented by discussing one that seems to have commended itself to some legal historians. Thus, Huebner in rejecting the survivalist theory of the avunculate for Germanic law writes: . . . the special honor of the maternal-uucle may have been merely a con- sequence of the fact that the maternal kindred came, in time, to be considered along with the paternal, who were at first exclusively regarded; in other words, a consequence of the fact that the family's purely agnatic structure was re- placed by a cognatic organization. In this appearance of the idea of eognatic relationship, which transformed in the same manner the family and the sib . . . , the maternal uncle naturally played the most important role: he was the link between the families of the father and the mother, and he was primarily the person upon whom was incumbent, as the representative of the maternal sib, the protection of the wife as against her husband. so My comment on this would simply be that it is unnecessary to assume the seqiience from agnatic to eognatic institutions : matronymy is perfectly consistent with the assignment of definite functions to the father 's group and patronymy is equally consistent with the avunculate. This point of view, combined with transmission and the influence of cross-cousin marriage, accounts in my opinion, for the vast majority of recorded avuncular institutions, though I am (luite willing to admit that there is a slightly greater probability for the avunculate to be coupled with matronymy than with patronymy. The case might be favorable for a higher degree of correlation if we could disengage instances of borrowing from those where the custom has sprung up R. Huebner, History of Germanic Private Law, p. 590, 1018. 191i)] Lowie: The Malrilincal Complex 43 spontaneously, bnt this we are unfortnnately not able to do except by specixlation. Yet even so, the correlation wonld prove more involved than if the avnnculate were simply a corollary of the matronymic institution. That is to say, it is not matronymic tribes, but matronymic tribes of a particular type, that seem to form a favorable soil for the evolution of avuncular customs. The absence of such customs among the Australians and their development in Africa, Melanesia, and the settled tribes of North America indicate that possibly there is a multiple correlation with matronymy and a settled mode of existence. On the other hand, it may turn out that matriloeal residence is also largel.y involved. In short, instead of saying that matrilineal societies tend to give rise to the avuncular usages, we may ultimately come to make the statement that the coefficient of correlation for the avuncu- late with sedentary tribes that are both matrilineal and matriloeal is .75; that the coefficient for nomadic matrilineal tribes is .05; while for nomadic patrilineal and patrilocal peoples it approximates zero. But these are merely suggestions thrown out to stimulate further research. It should be noted that the avunculate involves an interesting problem in diffusion. There are certainly very noteworthy resem- blances, for example, between the Banks Islands, the Tlingit, and the Bantu forms of the practice. If we assume with the extreme dif- fusionist school that no cultural feature can arise independently in two distinct parts of the globe, the matter is very simple. In that case we should postulate that the avunculate developed once among tlie Banks Islanders, for instance, and was thence transmitted to Africa and America and wherever else its observed range of distri- bution may extend. For all we could tell its origin would be an accidental occurrence since ex hypothesi it represents a unique phenom- enon. No matter what may have been its concomitants we are in no position to manipulate them so as to separate factors that helped from those which hindered its evolution. Any suggestion as to causal connections would thus necessarily remain arbitrary, that is, unamen- able to any mode of verification. The matter stands very differently if we accept the view current in America that similar cultural features may arise independently in unconnected areas. In this case an irrepressible logical instinct leads us to posit like conditions as underlying like observed effects. The similarity of avuncular usages in Melanesia and North America then appears as the probable, if not inevitable, consequence of like con- lyffii&iim.^^ja-'- ' A^jM^j.yfe^^^ 44 University of California Puolications in Am. Arch, and Ethn. [\ ol. 16 eomitant eireumstanees, and it becomes the duty of the ethnologist to ascertain what are the significant concomitants. If the avuncnlate is, mathematical!}^ speaking, a function of a series of features includ- ing matronymj', its occurrence in geographically and historically distinct communities ceases to puzzle, provided the same correlates are alwaj's associated with it. Practically the matter would .stand thus. Independent development would be postulated for the discon- nected areas of the globe. These would not be determined once and for all time by abstract geographical speculation, nor by general cultural considerations, but until exclusive reference to the one trait under discussion. For example, in a certain sense all of the New World forms a cultural unit. But this fact is negligible for the aAainculate when we find it among the Haida on the one hand and the Hopi on the other. There is no possible way to account for the absence of the custom in the immense intervening area except to assume that it never existed there. In other words, the Hopi avuncu- late represents one independent evolution, the Northwest Coast parallel another. "When such primary centers become foci for the transmission of the avunculate witliout at the same time transmitting the correlated traits, we are likely to find the observed facts of dis- tribution — great resemblance between disconnected groups sharing certain features besides the avunculate, and the occurrence of the avunculate in other localities which lack all the essential correlates for the independent evolution of the avunculate, but are in geographical proximity to localities that have developed it. To return to the general problem for a summary of results based on an empirical survey. The theory of a matrilineal society which by some necessity produces out of itself a series of features whose presence in turn may be used to establish the existence of such a society in the present or past is untenable. It ignores two vital groups of empirical phenomena — the frequent absence of the supposed symptoms among undoubtedly matrilineal peoples, and the enormous extent of borrowing, which accounts far more satisfactorily than the survival hypothesis for the occurrence of the avunculate amidst patrilineal institutions. Some degree of correlation between matronymy and matrilocalism or the avunculate may be accepted, but everything points to the conclusion that the connection is a far more intricate one than is commonly supposed. Here again discrimination is a prerequisite to a sane envisaging of the problem. The degree of correlation need not be the same for all of the supposed constituents 1919] Lowie: The Malrilineal Coinplex 45 of the matrilineal complex; in fact, all probability is to the contrary. For those with whom the a priori plausibility of the matrilineal com- plex theory in its classical form still weighs heavily, a brief historical retrospect is recommended. The earliest theoretical treatise on matronymy interpreted the feature as a sign of the matriarcJiate. Nothing could have been more plausible ; for what more naturally accounts for matrilineal descent than female ascendancy ? Yet in the face of a truly overwhelming mass of negative evidence the followers of Bachofen have long ago abandoned the conception of the matri- archate as a necessary or even common correlate of matronymy. In proportion as supposedly matriarchal tribes have become better known, the "mother-rule" has shrunk into certain property rights held by women (Hopi, Khasi), or certain social and political prerogatives (Iroquois). The correlation of these comparatively meager privileges with matronymy may possibly be expressed b^' a coefficient of .01, though I seriously question whether it is nearh^ so strong. A priori reasonableness can accordingly not take the place of empirical facts. Let us study what sociological traits are actually linked together, and we shall then have something to contribute to the problem of the matrilineal complex. UNIVERSITY OF CALIFORNIA PUBLICATIONS- (CONTINUED) Vol. 7. 1. The EmeryviUe Shellmoimd, by Max XJhle. Pp. 1-106, plates 1-12, with 38 text figures. June, 1907 1.25 2. Eecent Investigations hearing upon the Question of the Occurrence of Neocene Man in the Auriferous Gravels of California, by William J. Sinclair. Pp. 107-130, plates 13-14. February, 1908 35 3. Porno Indian Basketry, by S. A. Barrett. Pp. 133-306, plates 15-30, 231 text figures. December, 1908 1.75 4. Shellmounds of the San Francisco Bay Region, by N. C. Nelson. Pp. 309- 356, plates 32-34. December, 1909 50 5. The Ellis Landing Shellmoimd, by N. 0. Nelson. Pp. 357-426, plates 36-50. April, 1910 75 Index, pp. 427-443. Vol.8. 1. A Mission Record of the California Indians, from a Manuscript in the Bancroft Library, by A. L. EJroeber. Pp. 1-27. May, 1908 25 2. The Ethnography of the CahuUla Indians, by A. L. Kroeber. Pp. 29-68, plates 1-15. July, 1908 ,75 3. The Religion of the Luisefio and Diegueno Indians of Southern California, by Constance Goddard Dubois. Pp. 69-186, plates 16-19. June, 1908 1.25 4. The Culture of the Luisefio Indians, by Philip Stedman Sparkman. Pp. 187- 234, plate 20. August, 1908 50 5. Notes on Shoshonean Dialects of Southern CaUfomia, by A. L. Kroeber. Pp. 235-269. September, 1909 35 6. The Religious Practices of the Diegueno Indians, by T. T. Waterman, Pp. 271-358, plates 21-28. March, 1910 80 Index, pp. 359-369, Vol.9. 1. Yana Texts, by Edward Sapir, together with Yana Myths collected by Roland B. Dixon. Pp. 1-235. February, 1910 2.50 2. The Chumash and Costanoan Languages, by A, L. Kroeber. Pp. 237-271. November, 1910 - .35 3. The Languages of the Coast of California North of San Francisco, by A. L. Kroeber. Pp. 273-435, and map, April, 1911 1.50 Index, pp. 437-439. Vol.10. 1. Phonetic Constituents of the Native Languages of California, by A. L. Kroeber. Pp. 1-12. May, 1911 10 2. The Phonetic Elements of the Northern Paiute Language, by T. T. Water- man. Pp. 13-44, plates 1-5. November, 1911 45 3. Phonetic Elements of the Mohave Language, by A. L. Kroeber, Pp. 45-96, plates 6-20. November, 1911 „ 65 4. The Ethnology of the SaUnan Indians, by J. Alden Mason. Pp. 97-240, plates 21-37. December, 1912 1.75 5. Papago Verb Stems, by Juan Dolores. Pp. 241-263. August, 1913 25 6. Notes on the Chilula Indians of Northwestern California, by Pliny Earle Goddard. Pp. 265-288, plates 38-41, April, 1914 30 7. Chilula Texts, by Pliny Earle Goddard. Pp. 289-379. November, 1914 1.00 Index, pp. 381-385. Vol. 11. 1. Elements of the Kato Language, by Pliny Earle Goddard, Pp. 1-176, plates 1-45. October, 1912 2.00 2. Phonetic Elements of the Diegueno Language, by A. L. Kroeber and J. P, Harrington. Pp. 177-188. April, 1914 10 3. Sarsi Texts, by Pliny Earle Goddard. Pp. 189-277. February, 1915 1.00 4. Serian, Tequistlatecan, and Hokan, by A. L. Kroeber. Pp. 279-290. Febru- ary, 1915 10 5. Dichotomous Social Organization in South Central California, by Edward Winslow Gifford. Pp. 291-296. February, 1916 ..._ 05 6. The Delineation of the Day-Signs in the Aztec Manuscripts, by T. T. Water- man. Pp. 297-398, March, 1916 1.0& 7. The Mutsun Dialect of Costanoan Based on the Vocabulary of De la Cuesta, by J, Alden Mason. Pp. 399-472. March, 1916 70 Index, pp. 473-479. glljjj I CONGRESS 019 ssi^TTel UNIVERSITY OF CALIFORNIA PUBLICATIONS- (CONTINUED) Vol. 12. 1. Composition of California Shellmounds, by Edward Winslow Gifford. Pp. 1-29. February, 1916 80 2. Callfomia Place Names of Indian Origin, by A. L. Kroeber. Pp. 31-69. June, 1916 40 3. Arapaho Dialects, by A. L. Kroeber. Pp. 71-138. June, 1916 70 4. Miwok Moieties, by Edward Winslow Gifford. Pp. 139-194. June, 1916 55 5. On Plotting the Inflections of the Voice, by Cornelius B. Bradley. Pp. 195- 218, plates 1-5. October, 1916 25 6. Tiibatulabal and Kawaiisu Kinship Terms, by Edward Winslow Gifford. Pp. 219-248. February, 1917 30 7. Bandolier's Contribution to the Study of Ancient Mexican Social Organiza- tion, by T. T. Waterman. Pp. 249-282. February, 1917 35 8. Miwok Myths, by Edward Winslow Gifford. Pp. 283-338, plate 6. May, 1917 55 9. CaUfomia Kinship Systems, A. L. Kroeber. Pp. 339-396. May, 1917 60 10, Ceremonies of the Pomo Indians, by S. A. Barrett. Pp. 397-441, 8 text- figures. July, 1917 •. .45 11. Pomo Bear Doctors, by S. A. Barrett. Pp. 443-465, plate 7. July, 1917 25 Index, pp. 467-473. Vol. 13. 1. The Position of Yana in the Hokan Stock, by E. Sapir. Pp. 1-34. July, 1917 35 2. The Yana Indians, by T. T, Waterman. Pp. 35-102, plates 1-20. February, 1918 75 3. Yahl Archery, by Saxton T. Pope. Pp. 103-152, plates 21-37. March, 1918 .75 4. Yana Terms of Eelatlonship, by Edward Sapir. Pp. 153-173. March, 1918 .25 Vol. 14. 1. The Language of the Salinan Indians, by J. Alden Mason. Pp. 1-154. January, 1918 1.75 2. Clans and Moieties in Southern Callfomia, by Edward Winslow Gifford. Pp. 155-219, 1 figure in text. March, 1918 75 3. Ethnogeography and Archaeology of the Wiyot Territory, by Llewellyn L. Loud. Pp. 221-436, plates 1-21, 15 text-figures. December, 1918 2.50 4. The Wlntun Hesi Ceremony, by S. A. Barrett. Pp. 437-488, plates 22-23, 3 figures In text. March, 1919 75 Vol. 15. 1. Ifugao Law, by R. F. Barton. Pp. 1-186, plates 1-33. February, 1919 2.00 Vol. 16. 1. Myths of the Southern Sierra Miwok, by S. A. Barrett. Pp. 1-28. March, 1919 30 2. The Matrilineal Complex, by Robert H. Lowie. Pp. 29-45. March, 1919 15 Volumes now completed: Volume 1. 1903-1904, 378 pages and 30 plates $4.25 Volume 2. 1904-1907, 393 pages and 21 plates 3.50 Volume 3. 1905. The Morphology of the Hupa Language, 344 pages 3.50 Volume 4. 1906-1907. 374 pages, with 5 tables, 10 plates, and map 3.50 Volume 5. 1907-1910. 384 pages, with 25 plates .^ 3.50 Volume 6, 1908. 400 pages, with 3 maps 3.50 Volume 7. 1907-1910. 443 pages and 50 plates 3.50 Volume 8. 1908-1910. 369 pages and 28 plates 3.50 Volume 9. 1910-1911. 439 pages 3.50 Volume 10. 1911-1914. 385 pages and 41 plates 3.50 Volume 11. 1911-1916. 479 pages and 45 plates 3.50 Volume 12. 1916-1917. 473 pages and 7 plates 5.00 Note: — The University of California Publications are offered in exchange for the publi- cations of learned societies and institutions, universities, and libraries. Complete lists of all the publications of the University will be sent upon request. For sample cop ies, l ists of pubUcations or other information, address the MANAGER OF THE UNIVERSITY PRESS, BERKELEY, CALIFORNIA, U, S. A. All matter sent in exchange should be addressed to THE EXCHANGE DEPARTMENT, UNIVERSITY LIBRARY, BERKELEY, CALIFORNIA, U. S, A.