DENSITIES OF OXYGEN AND HYDROGEN 



SERIES FOURTH. 



1 Oct'r 19 282.7 21.6 43.9 40.7 41.6 42.1 1 II 42.1934 704-43 3-517' I089972] 



2 23 249.9 21.0 53.8 50.0 41.6 42.1 II I 42.2027 742.83 3.7057 .089S77 



3 25 253.0 22.3 52.9 48.9 41.6 42.1111 42.2016 740.66 3.6944 .089867 



4 31 254.8 19.6 524 48.9 41.6 42.1111 42.2016 740.84 3.6973 .089916 



5 Nov'r 2 251.2 21.5 53.4 49.5 41.6 42. nil 422007 738.33 3.6790 1.089777] 



6 4 236.5 22.5 57.8 53.4 41.6 42. nil 42.2046 743.51 3.7080 .089846 



If we increa.se the niemis by <ine tliirty-thousaiidtlu* we have 



Series III., 1)^ - 0.089886 gr. ± 0.000004 9 

 Series IV., I),, ^ 0.089880 gr. ± O.OOOOO.s 8 



L*5. — iri'DROGEN BY NKW METHOD. SOURCES OF ERROR. 



If, after measuring the vacuum in the globes, leakage occunvd, tlie ])i'essure of 

 the hydrogen in the manometer would a[)j)ear too great, and the density, therefore, 

 too small. Now it is believed that in e.vperiment 5 of series iv. this accident occurred. 

 The key of stopcock e. Fig. '2-\, was found, afterwai'd.s, to have formed an in- 

 cipient crack at some previous and unknown time, and leakfige from tlii-< (•.■lusc had 

 become so trouljlesome tliat, unaV)le to detect the real source, and asci'ibing it to 

 the wrong source, I broke up tlie apparatus and made a new one in which leakage 

 was impossible, and in which it never took place. This crack was in such a ])osi- 

 tioii on the key that the leak was likely to be developed only when the sto]K'oek 



* Sec note, ]jage 28. 



