40 BUREAU OF AMERICAN ETHNOLOGY [Boll. 192 



On the whole, the trash mounds at Old Town are not unhke the 

 Tlingit middens of the Angoon area in southeastern Alaska (de 

 Laguna, 1960), or the larger middens explored by Drucker (1943) 

 in British Columbia. They differ, however, from middens in the 

 Pacific Eskimo-Aleut area in containing far more earth or sand, and 

 far less sheU. The proportion of artifacts in all middens of the 

 Northwest Coast is also much lower than in sites within Chugach, 

 Koniag, and Aleut territory. This last may be due in part to the 

 greater use of perishable wood for artifacts on the Northwest Coast 

 and also to the fact that bone rots quickly in the more acid, less 

 sheUy deposits. Other differences may be due to cultural factors, 

 such as greater rehance on fish and less on sheMsh on the Northwest 

 Coast, less carelessness in losing and discarding possessions (a trait 

 for which the Eskimo are noted), or greater neatness in disposing of 

 rubbish at village sites. Such care does not seem to have been taken 

 at temporary camps or forts of the Tlingit (de Laguna, 1953, p. 55). 

 Although our informants doubtless exaggerate the neatness of aborig- 

 inal housekeeping, the custom of frequently replacing the sand or 

 gravel on the floor around the fireplace, which had to be done in any 

 case before a shamanistic seance, might account for the high proportion 

 of sand in the middens. 



The concentration of fire-cracked rocks in the upper layers of the 

 deposits, in contrast to their relative scarcity in the lower layers 

 and in Mounds C and D, is paralleled at sites on Prince Wilham 

 Sound, Cook Inlet, and Kodiak Island. This has been interpreted 

 (de Laguna, 1934, p. 162; Hrdhcka, 1944, pp. 30, 133, 394; Heizer, 

 1956, pp. 23 f.; de Laguna, 1956, pp. 49, 266) as indicating the rela- 

 tively late appearance among the Pacific Eskimo of the steam bath. 

 This type of bath has a more limited distribution and so is presumably 

 more recent than sweat bathing in heated air without steam. It 

 also necessitates far greater use of hot rocks than does stone boiling 

 of food. The steam bath seems to be older on Prince William Sound 

 than on Cook Inlet and Kodiak Island, w^hich suggests diffusion from 

 the Chugach or the Northwest Coast Indians. 



No clear proof of this hypothesis has yet been established. Thus 

 Drucker (1943) noted no great concentration of fire-cracked rocks 

 in the Tsimshian and Kwakiutl middens he explored, although such 

 rocks were common and appear to have been fairly evenly distributed 

 through the middens. We do not know, however, if there were 

 so many as to suggest the steam bath. Ethnologically, sweat houses 

 on the Northwest Coast are confined to the Tsimshian, Haida, and 

 Thngit, although the steam bath was also taken by the northern 

 Kwakiutl (Drucker, 1950, Traits 375 and 667). Fire-cracked rocks 

 are common in Thngit sites in the Angoon area. Although Birket- 



