154 DROSEKA ROTUNDIFOLIA. CHAP. VII. 



given, which are so surprising that their credi- 

 bility requires every kind of support. In 1872 I 

 experimented on twelve immersed leaves, giving each 

 only ten minims of a solution ; but this was a bad 

 method, for so small a quantity hardly covered them. 

 None of these experiments will, therefore, be given, 

 though they indicate that excessively minute doses 

 are efficient. When I read over my notes, in 1873, 

 I entirely disbelieved them, and determined to make 

 another set of experiments with scrupulous care, on 

 the same plan as those made with the nitrate ; namely 

 by placing leaves in watch-glasses, and pouring over 

 each thirty minims of the solution under trial, treat- 

 ing at the same time and in the same manner other 

 leaves with the distilled water used in making the 

 solutions. During 1873, seventy-one leaves were thus 

 tried in solutions of various strengths, and the same 

 number in water. Notwithstanding the care taken 

 and the number of the trials made, when in the 

 following year I looked merely at the results, without- 

 reading over my observations, I again thought that 

 there must have been some error, and thirty-five fresh 

 trials were made with the weakest solution ; but 

 the results were as plainly marked as before. Al- 

 together, 106 carefully selected leaves were tried, 

 both in water and in solutions of the phosphate. 

 Hence, after the most anxious consideration, I can 

 entertain no doubt of the substantial accuracy of my 

 results. 



Before giving my experiments, it may be well to premise that 

 crystallised phosphate of ammonia, such as I used, contains 

 35 - 33 per cent, of water of crystallisation ; so that in all the 

 following trials the efficient elements formed only 64 - 67 per 

 cent, of the salt used. 



Extremely minute particles of the dry phosphate were placed 



