AND RATIO OF THEIR ATOMIC WEIGHTS. (39 



15. HYnilOCKN BY NEW MKTIKID. EXHAUSTION OF (iLOlU'iS. 



When tlie capacity of the coiiiicctiiiy tubes liail been (leteniiiiRHl, the stoj)- 

 cocks of the globes were opened, the pump was set in action, and a sufficient 

 exhaustion obtained and measured with the McLeod gauge wliicli was mentioned 

 on page 18. The measurement was not made till about halt' an liour after tin; 

 pump had been stopped. The automatic pumj) by acting during one night wouhl 

 produce a vacuum of ten, twenty, or forty millionths, according to circumstances; 

 when acting through one day and two nights, it would leave from two millionths 

 to three ten-millionths of an atmospliere. But such a degree of exhaustion was 

 unneces.sary in experiments in whicli simply the inci'ease in pressure due to the 

 admission of a known weight of hydiogen was to be determined. 



16. — HYDKOGEN BY NVAV METHOD. MANIPULATION OK TUBE CONTAINING PALLADIUM. 



It was thought to be of prime im})ortance to use on the tube containing palla- 

 dium no stopcock requiring lubrication or admitting the possibility of leakage. 

 It is necessary to determine the volume of this tube before and after each experi- 

 ment. If the lubrication of the stopcock is exposed to water during the bydi'o- 

 static weighings, a source of uncertain ei-ror is inti'oduced. It would be possible 

 to wait till the effect of this exposure should l)e tliought to have passed off, or to 

 immerse the palladium tube only to a certain mark in the hydrostatic weighing. 

 But the manipulation without any stopcock is easy and more accurate, and was 

 always used in the ex])eriments of this series, and in all the following experiments 

 by this method. 



A strong argument in favor of dispensing with the stopcock is the difficulty of 

 knowing that, in a given case, the stopcock did not leak. A sto})cock is often 

 found, after an expej-iinent, not to have leaked. I have lepeatedly exhausted a 

 globe, weighed it, connected it again to the pump, exhausted the coiuiectiug tubes to 

 the same degree as before, and then have opened the globe and again measured the 

 vacuum. When the lubrication was fi-eshly applied, not infrequently the amount 

 of leakage in one, two, or even four days lias been found to be negligilde. But 

 these stopcocks were of very special dimensions ordei-ed for the purpose, and even 

 these often leaked, causing great annoyance and loss of labor. If some such meas- 

 urement, or other proof that leakage has not taken place, cannot be applied, it 

 would be necessary to use some method which should entirely eliminate leakage. 



The tube holding the palladium was therefore arranged as shown in Fig. 25. 

 At d is the tube by which hydrogen is admitted ; at e is a drying tul)e, filled with 

 phosphorus peutoxide; at ^ is a plug of fusible metal, at h is a small bulb filled 



