•)4 DENSITIES OF OXYGEN AND HYIJKOOEN 



It'iistli ; liiv isfiile aud tlu" leii<,ftli to be iiieasiirecl were as near as tluy loiilel Ije <ni 

 any i)lau except tliat lately used by Lord Rayleigli, in wliirli the scale was in the 

 bai-ometer. Fiirthei-, the adjustiiK-iit of the level of the telescope of a cathetoraeter 

 of the first mentioned type demands care and time ; but when the scale and l)arome- 

 ter are in contact, the error due to lack of lioiizontality in the line of collimation 

 of the telescope is lessened in the same ratio as that in which the distance of the 

 scale from the object measured is lessened. To reach a given degree of precision 

 re(piires much less care in adjustment an<l time is saved. 



It is obvious that if a case containing water is used so as to make certain of 

 the temperature of the column of mercury and of the scale, it is necessary to [dace 

 the barometer and its scale in contact to avoid eirois due to refraction at curved 

 parts of the plates of glass enclosing the water. 



To the column '/ can be clamped two pieces e and /' wlii«di carry two slides 

 moving vertically by means of the screws g and //. To these slides are fixed ways 

 on which the microscopes h, I, move liori/ontally l)y means of a rack and [linioii. 



The method of observation was as follows : The microscopes were made to 

 give distinct vision of the scales, and were then adjusted vertically till the terminal 

 divisions of the eye-piece micrometers coincided with certain divisions of the scale. 

 The microscoi)es were then moved by the focussing movement so as to give dis- 

 tinct vision of the meniscus of the mercury in the two tubes. The divisions of the 

 eye-piece micrometer were then virtually the divisions of the long scale carried foi- 

 ward into the centre of the tube. Tlie divisions of the mici'onieter corresponded 

 to twentieths of a millimetre; by proper manipulation of the illumination, it was 

 not ditticult to read the level of the mercury with a mean eiior less than one hun- 

 dredth of a millimetre, which is of course much moie than the precision attained in 

 other parts of the manipulation. 



It will be seen that in this method of ol)ser\ atioii ii is assumed that the two 

 slides on which the two microscopes move horizontally, are not horizontal, but 

 parallel; or, rather, that the deviation from parallelism is negligible in the distance 

 between the scale and the axis of the barometric tube. The accuracy of this as- 

 sumption is easily tested by exchanging the positions of the microscopes, or by 

 setting a scale ]terpendicularly before tlic microscopes, adjusting the micrometers 

 to coincide with certain <livisions of the scale, and then moving the scale farther 

 from the colnmii, and sc»-iug whether the divisions of the micrometer are again at 

 the samedistance a^iart. In iliis way, it was found safe to trust the nieidianism of 

 my reading microscopes to the hundicdih of a millimetre. 



The mercury used in tilling the tul)e of this barometer was puiitied by treat- 

 ment with nitric acid, and subsequent distillation in a \aeuum ; and these opera- 



