12 DRIFTINK SAND. 



g-ravels in Ijmbo- light enough to be rolled uj) a gentle 

 slope, and to be moYed without anv undermining taking- 

 place. Eleven samples of such rolling drift sand, as it may 

 be called, have b?en collected and examined (Tab. 11). 

 Its composition is much more uniform and regular than 

 that of lag gravels. The proportions of the different 

 grades arrange themselves in all the samples in two de- 

 creasing series on either side of a maximum, which in 

 three cases consists of coarse sand and in eight cases of 

 medium sand. In one sample the maximum grade consti- 

 tutes eighty-five per cent of the whole sample. The small- 

 est maximum is thirty-four per cent. All the maxima. 

 average fifty per cent in the eleven samples. Ninety per 

 cent of each sample is distributed among only three 

 grades in nine of the samples and among* four grades in 

 the other two. By different sampling, no doubt some- 

 what different results might be obtained. But these an- 

 ah'ses indicate that there is a rapid increase in the power 

 of the winds to roll quartz grains, when these begin to be 

 less than one millimeter in diameter. The same is also 

 indicated by the sudden decrease in several of the analyses 

 of lag gravels in the percentages of the grades, when this 

 limit of size is pa.ssed. (See analyses no. 1, 2, 3, 4, 5, and 

 7) . The rock fragments which exceed one millimeter in 

 diameter, are too large to be rolled uy) the i-ear slope af 

 a dune and are left in the '-blow-outs"" as a characteristic 

 ingredient in the lag gravels, but they ai'e, as may be 

 seen in the analyses, only very sparcely mingled in the 

 sand which is rolled up the rear slox)e of a sand drift. 



The sand which constitutes the main body of dunes 

 has been found remarkablv uniform in its mechanical 



