1909.] Vegetable Assimilation attd Resjnration. 5 



substance, and this might be quite different from the composition of the 

 true assimilatory increase. Experimental errors also are not negligible. 



If, with all these disturbing factors, a close agreement holds between 

 the observed increase of dry weight and the values calculated from the 

 results of the carbon analyses, it is clear that fixation of water cannot play 

 an appreciable part, still less produce an overwhelming positive error over 

 and above the real assimilatory gain. 



Errors from incomplete drying may be for the present regarded as 

 negligible. They are considered in Section VII (p. 45), where the pro- 

 cedure finally adopted for drying the very hygroscopic material is described. 



The following tables give the results of the combustion analyses and the 

 dry-weight results with which they are to be compared : — 



In Tables I to III the calculated value for the gain of dry weight has been obtained 

 by finding the weight of starch which is ecjuivalent to the observed gain of carbon, and 

 adding to this the observed increase in the ash content. 



As the ash consists largely of carbonate, a few specimens were analysed, using lead 

 chromate to decompose the carbonates of the alkali metals not decomposed by heat. 

 Tables IV and V give the results obtained. The calculated value for the increase of dry 

 weight is there simply the starch equivalent of the observed gain of carbon. 



The essentials of the conditions under which the material analysed was obtained are 

 given above each table. 



The examples of Heliantkus tuherosm in Table I were carefully powdered before the 

 final drying and combustion. The results in the other tables were obtained with material 

 which had been dried in a form convenient for immediate analysis.* 



The entire half-leaves were employed for the expeiiments in Table I and their area 

 found by applying a planimeter to photographic prints ;t the results are therefore all 

 calculated for an area of one square decimetre. 



Where equal areas from each half-leaf were cut, by mean.s of templets,t such as were 

 used by Sachs, or by the rotating punchf described on p. 42, no such calculation was 

 necessary, and for convenience in estimating the degree of accuracy the results have 

 been left in terms of the actual area used. This area is given for each experiment. 



Where the punch method was used the number of pieces punched from each half-leaf 

 is given ; each piece measured 0"804 square centimetre. In one or two cases the numbers 

 of discs from the two halves were not identical, and in these examples the figures for one 

 half are calculated for the number of discs used in the other half. 



* See p. 45. 



t The various methods of area determination are described in Section VII. For brevity 

 they will be referred to as the planimeter, templet, and punch methods respectively. 



