47 



weather prevented surveying work, we would muster six or eight men with 

 picks and shovels, clad in storm-proof rubher-coats, boots, and sou'westers, 

 and attack a shell-heap. Having, if possible, detected the kashini, one party 

 would enter the i)it which represented it, and dig away the embankments 

 from the inside, having first cleared away the superficial covering of vege- 

 table mold, often a foot deep, and the rank herbage upon it. This gave 

 them a good "face" to work on, and was the easier part of the work. The 

 othei's would start near the edge of the shell-heap, if possible taking a steep 

 bank bordering on the sea or on some adjacent rivulet, and run a ditch into 

 the deposit, going down until the primeval clay or stony soil was reached, 

 and this was steadily pushed, even when quite barren of results in the shape 

 of implements, until the day's work was done. This latter gave us a clear 

 idea of the formation and constitution of the shell-heaps ; enabled me to 

 distinguish between the different strata and their contents ; to make the 

 observations repeatedly; to fully confirm them by experience in many 

 localities; and thus to lay the foundation for the generalizations suggested 

 in this paper. While this work was barren in "finds" compared with the 

 excavations in the suj^erior and more modern accumulations, implements and 

 utensils were by no means entirely wanting; on the contrary, several hun- 

 dreds were collected in the period from 1871 to 1874, though I do not doubt 

 that we moved half a ton of debris for every specimen found. Thirty 

 specimens from all sources we considered a good day's woi'k, though we 

 frequently obtained a larger number and often fewer. We excavated in this 

 manner in Attn, Amchitka, Adakh, Atka, many localities in Unalashka, 

 Amaknak Island, and the Shumagins, and made casual examinations or slight 

 excavations in numerous other localities. 



In order to give a clearer idea of the arrangements of the village-sites, 

 I subjoin a sketch, not representing with exactness any special site, but 

 not dissimilar to one examined at Constantine Harbor, Amchitka. This 

 represents the outlines of the houses as more distinct than they are in reality. 

 The village had been built at the top of a steep bank, overlooking the broad 

 sandy beach of the harbor, and a small stream divided the base of the bank 

 from a marsh to the north of it. 



The absence of any differentiation into stone, iron, and bronze ages in 



