194 



Appendices to Fourth Annual Report 



As quickly as possible water was drawn off as above, until the lower 

 edge of the meniscus coincided with the mark. The tube was then lowered, 

 and allowed to remain for a few minutes lowered in the ice and water, 

 when if, as was almost invariably the case, the level of the water remained 

 constant it was taken out and suspended for ten minutes in pure water 

 having the same temperature as the room, then carefully wiped dry and 

 allowed to stand in the balance-room for an hour before weighing, the 

 open ends being loosely closed with small glass caps. 



The whole series of operations was then repeated, and if two weighings 

 agreed to within 0*5 mgr., the mean of the two, after deducting the 

 weight of the tube, was taken as the uncorrected weight of distilled water. 

 With proper care the difference between any two such weighings was 

 well within the above limit. When the difference was greater than 0*5 

 mgr. the determination was repeated. In such cases the difference was 

 almost always much greater than 0*5 mgr. ; varying from about 0*8 mgr. 

 to 2 or 3 mgr., and was easily traceable to the following source of error. 

 If the neck of the small cap be too narrow, it is not easy to avoid touch- 

 ing the glass with the end of the capillary terminal, in which case some 

 of the water flows out by surface attraction, and spreading over the inside 

 and along the neck of the cap gives rise to considerable loss by evapora- 

 tion. On one occasion differences greater than 0*5 mgr. were observed, 

 and traced to a want of care in keeping the temperature of the bath con- 

 stant. Such results were allowed no influence in determining the mean. 



The operations were exactly the same when the corresponding weights 

 of sea water were determined. The tubes were rinsed out with the water 

 in question not less than seven or eight times, using small quantities of 

 about 5 c.c, and allowing the last drop to drain away each time before 

 proceeding with the next rinsing. 



Before proceeding to the discussion of the results obtained in the above 

 manner, a few words are necessary on the balance and weights employed, 

 on the mode of weighing, and on the reduction of the weighings to 

 vacuo. 



It would be out of place to enter into a detailed description of the 

 balance which I used during the whole of this investigation. It is enough 

 to say, that it was specially constructed for me by Oertling of London on 

 lines laid down by Professor Dittmar, one of the first authorities on chemi- 

 cal balances. In an elaborate report upon this instrument, Professor 

 Dittmar found the delicacy of the balance, even when at the maximum, 

 to be practically independent of any charge up to 200 grammes, and such 

 that the probable error of a weighing executed in the ordinary manner 

 was only ±0'007 mgr. 



By taking the mean of six individual determinations as obtained by 

 seven readings of the excursions of the needle, the probable error was 

 reduced to ± 0*003 mgr. 



In summing up his report Professor Dittmar remarks : ' This is by far 

 ' the finest balance that ever passed through my hands.' 



The balance was repeatedly tested during the course of the investi- 

 gations, and the results fully corroborated this estimate. I gladly avail 

 myself of this opportunity to thank Professor Dittmar for the many 

 valuable suggestions he has from time to time made to me. 



The weights employed, also made by Oertling, were of platinum, and 

 were compared with each other, and the errors determined, before the com- 

 mencement and at the end of the investigation. On every occasion on 

 which the balance was used the zero point was redetermined, and in each 

 individual weighing not less than five excursions of the needle were noted, 

 not including the first, which was always neglected. I feel certain, there- 



