208 Composition of Dilute Sulphuric Acid. 



I chose that one of the four curves which appeared to me 

 (judging from Mr. Pickering's figures) to be based on ob- 

 servations which were especially free from abnormal errors. 



I chose that part of this curve which passed through the 

 percentage corresponding to the hydrate which Mr. Pickering 

 has isolated, and a range which is about one third of the 

 whole. As the result there are now extant three representa- 

 tions of the part of the curve which I studied : — 



(1) Mr. Pickering's drawings, which show four discon- 

 tinuities. 



(2) My curve, which Mr. Pickering frankly admits agrees 

 with the " experimental results just as satisfactorily as [his] 

 own drawings do " (p. 136). 



(3) Mr. Pickering has himself put forward a new solution 

 (footnote, p. 141) in which he divides the range into two 

 parts by means of two curves, which agree with my single 

 curve so closely that he himself states " that the magnitude of 

 the error will not help us to decide between their respective 

 merits." 



Now amid all the discussion as to whether the results are 

 best represented by continuous or discontinuous curves, one 

 thing is certain. Any peculiarity in a curve on which 

 argument is founded must be proved to be outside the error 

 of experiment. No weight can be attached to a representa- 

 tion which is not unique in the sense that no other is com- 

 patible with the observations when all allowance is made for 

 experimental error. Mr. Pickering's first solution (the four- 

 break one) is not unique. He has himself offered another 

 (one break) and I have contributed a third (no breaks), all of 

 which by his own admission are such " that the magnitude of the 

 error will not help us to decide between their respective merits.'''' 



Put briefly, the observations, accurate and numerous 

 as they are, are not accurate enough and not numerous 

 enough to decide whether the 18° density curve is or is not 

 discontinuous between 47 per cent, and 80 per cent. This is 

 what I claimed to have established, and Mr. Pickering has 

 joined forces with me by producing a second solution of his own. 



As to the deductions which are to be drawn from this 

 conclusion, 1 have previously expressed the opinion that Mr. 

 Pickering will enjoy a certain dialectical advantage until 

 some of his other curves have been examined in detail, and 

 that I do not intend to undertake that examination (which will 

 be lengthy and troublesome) myself. 



Absolutely convinced, as of course I am, of his bona fides, 

 I am not disposed to ignore his arguments as to the con- 

 cordance of his results, or to generalize hastily from my own 



