THE NEW SPECTRUM. 587 



and this law i.s found to hold o-ood in quantities so small that they 

 approach the physically inlinitesimal. In the actual bolometers. fJr 

 instance, the two arms of a Wheatstone bridge are formed of two 

 strips of platinum, side by side, one of which is exposed to the heat 

 and the other sheltered. The warming of the exposed one increases 

 its resistance and causes a deflection of the galvanometer. 



It was considered to be remarkal)le twenty years ago that a change 

 of temperature of one ten-thousandth of a degree Centigrade could be 

 registered; it is believed at present that with the consecutive improve- 

 ments of the original instrument and others, including those which 

 Mr. Abbot, of the Smithsonian Institution Observatoiy, has hitoly 

 introduced into its attendant galvanometer, less than one (me-hundr,'d- 

 miUionth of a degree in the change of temperature of the strip can 

 be registered. This indicates the sensitiveness of the instrument to 

 heat. 



As to the second relation. scM.ie measures have been made on the 

 steadiest light source obtainable. With ordinary photometric meas- 

 ures of its intensity one might expect a probable error of about 1 per 

 cent. The error with the bolometer was insensible by any means that 

 could be applied to test it. It is at an}" rate less than two one- 

 hundredths of 1 per cent. If we imagine an absolutely invisible spec- 

 trum, in which there nevertheless are interruptions of energy similar 

 to those which the eye shows us in the visible, then the bolometer, 

 whose sensitive strip passes over a dark line in the spectrum, visible 

 or invisible (since what is darkness to the eye is cold to it), gives a 

 deflection on the side of cold, and in the warmer interval between two 

 lines a deflection on the side of heat; these deflections being propor- 

 tionate to the cause, within the degree of accuracy just stated. 



The third quality, the accuracy of its measures of position, is better 

 seen by a comparison and a statement, for if we look back to the indi- 

 cations of the lower part of Lamansky's drawing we may see that at 

 least a considerable fraction of a degree of error must exist there in 

 such a vague delineation. Now, in contrast with this eaily record, the 

 bolometer has been brought to grope in the dark and to thus feel the 

 presence of narrow Fraunhofer-like lines by their cooler temperature 

 alone, with an error of the order of that in refined astronomical meas- 

 urement; that is to say, the probable error, in a mean of six observa- 

 tions of the relative position of one of these invisible lines, is less than 

 one second of arc: a statement which the astronomer, pr-rhaps, who 

 knows what an illusive thing a second of arc is. can best appreciate. 



The results of the writer's labors with the bolometer in the years 

 18S0 and 1881, and in part of his expedition in the latter year to .Mou.it 

 Whitney, were given at the Southampton meeting of the British .Asso- 

 ciation for the Advancement of Science in 1882.^ During th ese two 

 1 Eeport British Association, 1882. Mature xxvi, 1SS2. 



