- 
THE RARER ELEMENTS IN SOILS AND PLANTS. 25 
present escapes precipitation in both the sulphate and oxalate sepa- 
rations. Further, some strontium nitrate is probably dissolved by 
the comparatively large amount of absolute ether-alcohol which 
must be used to dissolve the calcium nitrate and completely wash 
it from the strontium nitrate. An error in the other direction arises 
through the difficulty of washing the strontium nitrate free from 
calcium salts. A spectroscopic examination of the final strontium 
precipitate with the Bunsen burner as a source of heat always showed 
the characteristic strontium line, and the calcium lines were very 
faint or missing. The writers believe the accuracy of the strontium 
determination to be such that the differences in plant composition 
shown in the tables for this element are not significant. 
SULPHUR. 
‘The sulphur determinations were made by two different methods, 
the Parr bomb method and the sodium-peroxide method. Both de- 
terminations yielded satisfactory results. 
CHLORINE. 
The determinations of this element were made according to Bulle- 
tin 107 (Revised) of the Bureau of Chemistry. Satisfactory results 
were obtained. 
PHOSPHORIC ACID, CALCIUM, AND MAGNESIUM. 
The phosphoric acid, calcium, and magnesium determinations were 
quite satisfactory, determinations by different individuals generally 
checking within 0.05 of 1 per cent. 
MOISTURE. 
The loss in weight in an electric oven kept at 105° C. was taken 
to represent the moisture. In some cases constant weight was reached 
by drying from four to six hours, but generally the drying had to 
run much longer. Volatile materials other than water were cer- 
tainly lost, as the strong continuous odor from the samples of pine 
needles and onions indicated. It was not practicable to dry the 
onions to constant weight, and the percentages are calculated on the 
air-dry basis. Throughout the work duplicate determinations were 
made until the analyst became very familiar. with the procedures, 
and thereafter difficult determinations were made in duplicate and 
unusual results were always checked up, generally by another analyst. 
