Henry Augustus Rowland. 255 



He was the first to point out the importance of reducing the results 

 from the scale of the mercurial to that of the air thermometer ; and for 

 this purpose he carried out a subsidiary investigation in which he com- 

 pared his mercury thermometers with an air thermometer. He found 

 this the most difficult part of the work, and he in fact attributed the 

 greater part of the probable error of his determinations to the uncer- 

 tainty of this reduction. His comparison of the thermometric scales 

 was however probably the most accurate achieved up to that date ; and 

 he gave as the result of this section of his paper a formula for the de- 

 viation of the mercury thermometer from the gas thermometer which 

 represents the scale-difference as accurately as it can be represented by 

 a simple type of expression. Subsequent corrections to this part of the 

 work, made by comparing his mercury thermometers with a Paris 

 standard mercury thermometer and also with a platinum thermometer, 

 have altered his original results by only one part in 2,000, which is no 

 more than the probable error of the result from other causes. 



His singular freedom from the defect common to experimentalists, 

 of over estimating the accuracy of their own work, is illustrated by the 

 estimate which he himself gave of the probable error of his determina- 

 tions. He considered that the value which he had obtained for this 

 fundamental constant was probably correct to at least one part in 500, 

 except near the extremity of the range of temperature over which he 

 worked where the great increase of the external heat-loss might have 

 produced a larger error. As a matter of fact his results near the 

 middle of his range agree to less than 1 in 2,000 with the most recent 

 determinations ; and although his values of the mechanical equivalent 

 appear to be in error by as much as 1 in 500 near the extremities of 

 his range, his values of the total Jieat of water, which was the quantity 

 directly measured in his method, are probably correct to 1 in 5,000 

 over the whole range. This achievement greatly surpassed in accuracy 

 all previous attempts in measuring a quantity of heat ; it was rendered 

 possible by the excellence of the mechanical design of his apparatus, 

 and by the precautions which he adopted to make the correction for 

 external heat-loss as small as possible and to determine it with accuracy. 

 These precautions enabled him to work over a range of temperature of 

 nearly 40C., as compared with a range of less than 4"C. employed by 

 Joule in his latest experiments. As a result he discovered for the first 

 time the remarkable fact that the specific heat of water suffers a notable 

 diminution, of nearly 1 per cent., between and 40C., which has been 

 amply verified by subsequent experimenters, in opposition to the work 

 of Eegnault, till then universally accepted, according to which the 

 specific heat increased continuously from to 100C. 



His next great work, that by which his name has become most 

 widely famous, was the invention and construction of the concave 



