38 Mr. D. Thoday. Experimental Researches on [June 11, 



beginning of an experiment. The stamping method described in Section VI 

 has been devised to fulfil this requirement. Tlie principle of the method is to 

 stamp each half of a leaf with a rectangle, by means of a specially constructed 

 rubber stamp, and to cut out with scissors the area so delimited. 



This method, considered merely as a means of determining area, is with 

 care probably as accurate as Sachs' templet method, and for many plants 

 must therefore supersede the latter. Further discussion of this new method 

 will be found in Section VI. 



In conclusion, it may be said without hesitation that errors from change of 

 area during experiment are by far the most serious and most difficult to deal 

 with of all the errors to which the Sachs dry-weight method in its original 

 form is liable. In consequence of the ignoring of these errors most of the 

 results which have hitherto been obtained with the method require revision 

 and repetition. 



The difficulties are not, however, insurmountable. Changes of area can 

 be estimated and corrected for, although the technique involved must be 

 laborious. On the other hand, the stamping method of area determination, 

 by' the use of which shrinkage errors are eliminated, is almost as simple in 

 use as Sachs' templet method, and it is hoped will prove widely applicable. 



Note. — Shrinkage errors affect any method which involves measuring the difference 

 between the amounts of a substance contained at different times or under different 

 conditions in unit area of leaf surface. The error is relatively great only when 

 the fluctuations are of a lower order of magnitude than the average content. 

 Thus, when increase in proteid coatent is to be measured, shrinkage errors might 

 be almost as serious as they are in measuring inciease of dry weight : on the other 

 hand, they could probably be ignored in determining changes in the amounts of starch 

 and other carbohydrates. 



A case which is j)robably an illustration of this iliti'eience in relative magnitude of 

 shrinkage errors is to be found in Men/.c's comparison of increase of dry weight with 

 gain of carbohydrates {loc. cit., p. •Vi). Although the conditions under wliich he 

 conducted his experiments were not such as would tend to produce large area changes, 

 his ash determinations favour the assumption that some .area changes did occiu- : if so, an 

 appreciable ])art of the excess increase of dry weight which is unexplained by the gain 

 of caibohydiates might be only <'ii>parent. On the other hand, some of the suiphis 

 probably rej)resents proteid formed, and the fluctuations in ash content may have been in 

 part or wholly real {cf. Section J 1 1, p. 9, footnote *). 



A better illustration is afforded by Saposchnikort''s results {loc. cit., see p. 10, ct s<'<j ). 

 He estimated jjroteids, as well iis starch and soliible carbohydrates : nevertheless, the 

 increase of dry weight was still not completely accounted for, and Ik; fell back on cellulose 

 as the possible form of the unexplained excess. It is much more ])robabk' that most of 

 the excess repi(^sents apparent increase of dry weight due to shrinkage. The higli rates 

 of increase indicatc(l by some of iiis shorter cxperinicnts favour this interpretation. 



