32 S. L. Penji< Id — Met hods for Determination of Water. 



the upper part of the tube from becoming- heated by radiation 

 from the flame, a screen of asbestus board can be used, or still 

 better a strip of wet cloth may be wrapped about the bulbs 

 and upper part of the tube, and this may be further cooled if 

 necessary by applying water from a wash bottle. After the 

 water has been expelled, the glass is fused down upon the sub- 

 stance, so as to make its volume as small as possible, and the 

 end containing the substance pulled off. It is not always nec- 

 essary to do this, but it usually is best. The tube containing 

 the water is wiped clean, cooled and weighed. The water can 

 be quickly removed by heating and aspirating out the steam and 

 moist air, when the tube after cooling is again weighed. To 

 test the accuracy of the method experiments were made with 

 transparent, colorless gypsum from Sicily. The mineral, 

 which is assumed to be pure and to contain 20*93 per cent of 

 water, was ground and weighed out air dry from a weighing 

 tube. The results are as follows : 



Gypsum taken. Water calculated. Water found. 



•20985 

 •2417 

 •2642 

 •3505 

 .4913 

 •6457 

 2-0000 

 5-0643 



•0439 

 •0506 

 •0553 

 •0734 

 •1028 

 •1351 

 •41S6 

 •0600 



•0442 

 '0505 

 •0553 

 •0731 

 •1028 

 •1352 

 •4195 

 •0623 



To these may be added some determinations 

 mine and natrolite, which by direct ignition in 

 7'76 and 9 '80 per cent of water respectively. 

 Calamine taken. Water calculated. Water found. 

 9. -5301 -0411 -0406 



Natrolite taken. 



10. -1838 -0180 -0176 



11. -2553 -0250 ■0247 



12. 1-0000 '0980 -0980 



Error. 

 •0003 + 

 •0001- 

 •0000 

 •0003 — 

 •0000 

 •0001 + 

 •0009+ 

 •0023 + 



made on cala- 

 a crucible lost 



Error. 

 •0005 - 



•0004 — 

 •0003 — 

 •0000 



These results indicate that the method is very accurate for 

 the determination of both large and small quantities of water 

 and experiments 7 and 8, which show the greatest apparent 

 variation, are within -045 and "04 per cent of the theory re- 

 spectively. 



To test the method where substances have to be added to 

 retain volatile products, experiments were made with pure 

 recry stall ized cupric and ferrous sulphates, containing 28*87 

 and 45*32 per cent of water respectively. To prevent the 

 escape of S0 3 some strongly ignited lime was mixed with the 

 powdered salts before heating. The mixing can best be done 



