136 MESSES. C. CUTHBERTSON AND E. PAKR METCALFE 



The standard density selected is that in which the number of atoms of the element 

 per unit volume is equal to the number of atoms contained in unit volume of hydrogen 

 at C. and 760 millims. 



In the present work the original procedure underwent several important improve- 

 ments. Repeated failure had shown that the glass tubes formerly used, with plate- 

 glass ends ground in and luted with shellac, were useless above 280 C., when the 

 shellac charred and the tubes cracked. 



Recourse was had, therefore, to tubes of fused silica, which were made with 

 admirable skill by the firm of HERAECS, of Hanau. The ends of the tubes, through 

 which the interfering rays passed, were plates of the same material ground optically 

 flat, fused into the tubes and again polished, so that the whole formed an air-tight, 

 homogeneous enclosure, which could be heated to the highest temperature employed 

 (about 850 C.) without fear of softening, and could be heated and cooled locally with 

 great rapidity without cracking. 



This property of the silica permitted the introduction of a second improvement in 

 the procedure. In the earlier work the two sections of the furnace were joined up in 

 the centre, and the whole length of the tubes was heated in one operation, so that the 

 observer had sometimes to remain with his eye at the telescope for several hours 

 while the furnace heated and cooled. This tedious method was now abolished. The 

 two sections of the furnace were separated by a gap of about an inch, and their inner 

 ends covered with thin asbestos boards. In these boards holes were punched, through 

 which the silica tubes passed. In this way the greater part of the tubes could be 

 raised to a high temperature while the gap in the middle remained comparatively 

 cool. With the object of obtaining a reasonable equality in the temperature of 

 the two halves of the furnace, the sections were made in all respects as nearly 

 similar as possible, and the heating coils were connected in parallel to the lighting 

 circuit. 



When the furnace had reached the temperature which was found by experiment to 

 be more than sufficiently high to evaporate the whole of the charge employed, cold 

 water was dropped on the exposed part of the tube until it was certain that the 

 temperature there must be below 100 C., and, consequently, that there could be 

 practically none of the element in a state of vapour. The observer then noted the 

 position of the bands in the telescope, in relation to a pointer fixed on the further 

 mirror, and his colleague quickly heated the central portion of the tube with a Bunsen 

 flame. The bands now moved rapidly and attained their maximum in a few minutes. 

 The heating was continued for about a minute after the stopping of the bauds 

 indicated the complete vaporisation of the charge ; the flame being then removed, a 

 second reading was obtained as the element condensed. When nearly all the bands 

 had passed, water was again dropped on the tube till it was certain that the zero had 

 been reached. Meanwhile the temperature of the rest of the furnace was kept 

 approximately constant, and thus many minor sources of error due to inequality of 



