240 



SCIENCE. 



[Vol. I., No. 9. 



;Woiild be subjected in use. It is this difficulty 

 of accui-ately knowing the temperature, that 

 has demanded the expenditure of so much time, 

 talent, and monej', in the construction of com- 

 pCiusating bars for base measurement. Were 

 it possible for a thermometer to accuratelj' 



register the temperature of a metallic bar 

 beside which it was laid, a simple rod or bar 

 would form the most accurate base-measuring 

 apparatus, as there w^ould be no risk of any 

 parts getting out of order. 



In the comparisons upon which he has been 

 engaged in the Bureau of U. S. standard 

 weights and measures, the writer has had 

 frequent occasion to notice the variations of 

 temperature under conditions wliich would 

 ordinarily be presumed to be the same ; and he 

 has had forcible evidence »f the fact, that no 

 matter how well the conditions are controlled, 

 or how carefuUj' the bars may be protected, 

 we can never rely upon two bars having the 

 same temperature. No doubt they often have 

 the same temperature ; or, when a difference 

 exists, its effect is inappreciable. Still, there 

 is an uncertainty in the matter which can be 

 eliminated only bj- a careful inter- 

 change of relative positions. 



A marked illustration of what is 

 said above is given in the following 

 case. The comparisons were made 

 by the writer. The circumstances of 

 comparison were as follows : seven 

 steel end-metres were to be com- 

 pared with a standard metre. The 

 comparisons were made in a room 

 about 20x16 feet in size. This 

 room was at the north-east corner of 

 the building, so that two of its sides 

 were outer walls. Thej' were about 

 two feet thick. The room was be- 

 low the level of the street in front, 

 but had free circulation of air around 

 the outer sides, the building being 

 separated from the coal-vaults in 

 front bj' an area five or six feet 

 wide. The comparator was parallel 

 to the eastern (longer) side of the room, and 

 about three feet from it. .The doors of the 

 room were kept closed, and the ds.\\y range of 

 the temperature seldom exceeded three degrees. 



Two windows in the eastern wall were closed 

 with a double glass sash and a solid wooden 

 frame ; and, to more effectually close them 

 against passage of air, heavy manila paper was 

 closelj' pasted over the entire frames. The 

 eight bars were supported on racks, in two 

 groups of four each, on both sides of 

 the position that a bar would have 

 when lying between the abutting- 

 screw and contact-slide of the com- 

 parator. As the bars lav in the 

 rack, the}' were about three-quarters 

 of an inch apart from centre to 

 centre. The extreme bars were about 

 seven inches apart. The arrangement is shown 

 in Fig. 1. 



The bed-plate of the comparator was a 

 framework of solid wood several inches thick. 

 The bars and comparator were covered by a 

 framework of wood and heavy plate-glass. 

 The manipulation of the bars was effected by 

 long pliers working through two narrow slits 

 in the top of the case. As an extra precau- 

 tion, a covering of heavy paper was placed 

 over the whole. In manipulating the bars, the 

 observer stood between them and the wall ; so 

 that, if the heat from his bodj' succeeded in 

 penetrating the casing, the effect would be to 

 dbninish the variations observed, and not to 

 increase or to produce them. Numbering the 

 notches in the rack from 1 to 8, the former 

 being the nearest to the wall, the order of ar- 

 rangement and change was as follows : the 



standard remained constantly in No. 5 ; after 

 each set of comparisons, the bar in No. 8 was 

 put in No. 1 ; each other bar being moved 

 forward one space, except that in No. 4, which 



