RICHARDS AND LAMB. — SPECIFIC HEAT OF LIQUIDS. 671 



For the specific-gravity deterininatious a thermostat was constructed 

 which maintained a constant temperature of 2o° within the one one-hun- 

 dredth of a degree. It was heated by means of an incandescent electric- 

 light bulb, and regulated by a simple regulator filled with alcohol and a 

 little mercury, which, instead of cutting off a current of gas, made and 

 broke an electric relay current, which in turn opened or closed the light- 

 ing circuit. A glass float was then made which was adjusted so that it 

 neither sank nor floated in a known mixture of acetone and water con- 

 taining a slightly greater percentage of water than was usually present 

 in the unknown mixtures. After each mixture had remained an hour in 

 the above thermostat the float was put in and water added from a burette 

 until the float just sank. Acetone was then added until equilibrium was 

 reached. Since the mixture to which the float was adjusted contained 

 about 3 per cent of water, evidently the volume of the water needed to 

 be known with much greater accuracy than that of acetone. The water 

 burette was therefore very narrow, and was graduated to hundredtiis of 

 a millilitre. The float was easily sensitive to less than one one-hundredth 

 of a gram of water, and since the total amount of water usually amounted 

 to somewhat more than 170 grams, the error in this method affected the 

 final result by less than one one-hundredth of a per cent. 



Wlien the solutes in the solutions whose specific heats are being deter- 

 mined are insoluble in acetone, this method can be used for them also. 

 This was the case with copper sulphate. If a given solute is soluble in 

 acetone the pipettes can be rinsed with water, and the amount of the 

 solute in this determined by chemical methods. This was the case with 

 silver nitrate, for example. 



Beckmann thermometers were used to measure the temperature of the 

 hot liquid, the cold liquid, and the mixture. The scales were all divided 

 into one-hundredths of a degree, and by means of a small lens could be 

 read to within a one-thousandth of a degree with certainty. A set of 

 Baudin thermometers, described elsewhere, was u.sed to standardize these.* 

 Comparisons were made in all cases under experimental conditions, and 

 the zero point of the standard was afterwards taken. 



The apparatus having been described, it is desirable to emphasize its 

 advantages. 



In the first place, all error from thermometric lag is avoided. Again, 

 such slight eriors from cooling as may be present during transference is 

 compensated by an equal and contrary heating effect, for while the liquids 



* Richards, Am. Jour. Sci. (4), 6, 20l (1898). 



