( -^^'4 ) 



indicated in Comm. N". 94 from the su|)i)]y bottle into the apparatus, 

 which tube is closed in other cases with a i-iibber lube with 

 cock. 3. an outflow tube </., (Iit>-. 2), which leads alonj^ cock k 

 (fig. 1 and fig. 4) to the gasholder with pure hydrogen, to a safety 

 tube / (fig, 1), along /:, to an airpump, and past k^ to the vacuum 

 bottle r, from which the li<juid hydrogen is siphoned over (the ope- 

 ration is elucidated by the diagrammatic fig. 4, which does not call 

 for a further description). 



We first have convinced ourselves tliat when the air has been 

 exhausted from the hydrogen tube surrounded by c, this tube exactly 

 occupies its place between the poles, witiiout being strained by the sup- 

 ports q and /, when these have a suitable position, we then fill it along k^ 

 with hydrogen from the gasholder, exhausting it repeatedly, then 

 we pour liquid air through a funnel with filter into 6i, which is 

 covered with some cotton wool. The apparatus is then fdled with 

 liquid hydrogen through d^. In order to pass to the melting 

 point of hydrogen, /(•, is opened lill crystals appear on the 

 surface of the liquid hydrogen, through which the gas bubbles 

 which rise from the heated crystal, are seen to make their waj'. If 

 the apparatus has been filled in the way described before, observations with 

 the crj'stals may be made uninterruptedly for several hours. The pre- 

 cautions taken to prevent mixing of hydrogen and air are indis- 

 pensable. Air entering the apparatus, would siiü^; down, and be sucked 

 up in front ot the crystal as soon as the magnetic field is applied, 

 and intercept the light. 



For every filling of the apparatus 74 li^®'" of liquid hydrogen from 

 the supply is generally used, and it was sufficient to do this twice 

 a day to be able to observe all the day in case of ordinary as 

 well as of low pressure; twice a week a quantity of 5 liters was pre- 

 pared for these experiments, which was Just sufficient to fill the appa- 

 ratus also the second day after the preparation. As it was impossible 

 to entirely prevent the hydrogen which evaporated at lowered pres- 

 sure from being contaminated with air, it was not admitted again into 

 the cycle. The hydrogen cycle proved its reliability by never failing 

 us a single time in all these weeks. 



I. Phenomena which depend solely on the temperature. 



§ 3 Siinplification of the spectra. On cooling to the temperature of 

 liquid air (7' =85''^ one of us had found ^) that almost all bands 

 become narrower and divide, some new ones also appearing. In 



'"I Jean Becquerel, 1. c. 



