16 MACGEEGOE ON THE DENSITY OF WEAK 



taiued could be read to O'OS c.c. The water, with which the bottles were thus Jilled, had 

 been freed from air uuder the receiver of au air-pump. 



To test the tightness of the stoppers, the dilatometers were filled, until the upper 

 surfaces of the water were near the tops of the tubes. The stoppers were thus subjected 

 to as great pressures as they would be during the experiments. After the bottles had 

 taken the temperature of the bath, I observed the variation of the height of the water in 

 the tubes from time to time, until I had satisfied myself that there was no leak, — a return 

 to a formerly observed height in one bottle being accompanied in all cases by a similar 

 return in the others. 



I next satisfied myself that differences of temperature between the bottles, greater 

 than any which could arise during the experiments through the dissolving of salt in some 

 bottles and not in others, would vanish in less than the time that was to intervene between 

 successive measurements. 



As the dilatometers could not be kept at constant temperature, and as any change of 

 volume of their contents must therefore be partially due to change of temperature, it was 

 necessary to know the relative apparent thermal expansion of their contents. For this 

 purpose, both at the outset when all the bottles contained water, and at intervals during 

 the series of experiments when some of them contained solutions, the temperature of the 

 bath was varied, and the heights of the water or solutions in the différent tubes were 

 observed when the bottles had assumed the temperature of the bath. These results were 

 tabulated for i^urposes of correction. 



The solutions, whose volumes were measured, were formed by the addition of known 

 masses of anhydrous salt to the water in the bottles. The salt was simply dropped little 

 by little down the tubes of the dilatometers. In some cases no difficulty was experienced ; 

 in others the salt was found to cake occasionally at the surface of the liquid. In these 

 cases various expedients were adopted to hasten the solution ; but the greatest care was 

 taken to prevent the loss, either of any of the salt which had been weighed out for solu- 

 tion, or of any of the liquid in the bottles. When the desired amount of salt had been added 

 to a bottle, the upper end of the tube was closed with a small cork to prevent evaporation, 

 and the bottle was put in the bath. After an interval of about twenty-two hours the bottle 

 was taken out, and if the salt was found to be dissolved, was first well corked and then 

 rolled, until its contents had been thoroughly mixed. It was then replaced in the bath and 

 left for another hour, when the height of the free surface of the liquid was observed. _ Not 

 possessing a cathetometer, I required, for measuring differences of level, to trust to a 

 steel scale placed in contact with the tube. Care was of course taken to avoid paral- 

 lactic errors as much as possible. 



To one of the four bottles no salt was added; and it was kept carefully corked up, so 

 that the quantity of water it contained might be constant. The variation of the height 

 of the water in the tube of this bottle was due, of course, to change of temperature alone. 

 This variation being observed, and the relative apparent thermal expansions of the liquids 

 in the four bottles being known from the subsidiary experiments referred to above, the 

 variations, due to changes of temperature, of the heights of the solutions in the tubes of 

 their respective bottles could be determined and eliminated. The variations of tempera- 

 ture were in all cases slight, the bath being large and its daily thermal history being very 

 constant. 



