niilier parts of the geuerator tn))es. indicating that at least one of the 

 suhstauces still contained moisture. 



Other portions of the same mixtures as above AA'ere dried in their tubes 

 for several hours at 100°-10o°. Chlorine was evolved upon heatiny to 

 122° Ijut no oxygen Avas evidenced l>.v a spark. At 13.")° the rate of evolu- 

 tion of oxygen was approximately in direct proportion to the amount of 

 manganese dioxide used, this lieing the reverse of the case when the ma- 

 terials were not dried. This, however, is not stated as a definite law. 



Four other mixtures were mofe carefully dried, then heated in the 

 bath. Chlorine was evolved at 14^)°, oxygen at 108°. 



It was early seen that no reliable i-esults could be obtained so long as 

 the manganese dioxide held moisture. To determine whether this sub- 

 .stance Avas hygroscopic, and if so. roughly the amount of water taken up, 

 some freshly dried material Avns AA-eighed in a closed bottle, then allowed 

 to stand open for definite periods, Aveighing after each period. In tAventy 

 minutes its Aveight increased approximately 1 per cent.: after one and a 

 half hours. .'I per cent.: after forty-five hours. <i per cent. 



To determine the difference in bthavior due to this moisture, Iaa-o mix- 

 tures AA-ere prepared: In (Xi the manganese dioxide AA^as dried oA-er a 

 free flame. Aveighed in a glass-stoppered bottle and the Aveighed potas- 

 sium chlorate added. The other mixture (Yj Avas of potassium chlorate 

 and ordinaiw undried maugauese dioxide: both were molecular mixtures. 

 In this and future experiments chlorine Avas tested for hj starch and 

 potassium iodide paper. At 125° (X) gave no chlorine or oxygen, (Y) gave 

 large quantities of chlorine but no oxygen. Much moLsture collected in 

 (Yi. At 148° a .steady stream of oxygen came from (Y). continuing as long 

 as heated. Xo trace of chlorine or oxygen came from (X). 



More manganese dioxide Avas purified by digesting in cold distilled 

 Avater, then washing until free from chlorides. The AA'a.sh water contained 

 small amounts of manganese and cal<-iuni. Tlie Avashed mass Avas dried 

 for tAvo and a half hours at 200°-21(r. 



Foiir tubes were noAv lilled Avith mixtures in molecular proportions, 

 transferring the manganese dioxide quickly at 211(1" to the hot Aveighing- 

 bottle, cooling, weighing, adding the ground and weighed potassium 

 chlorate, and mixing. The mixtures Avere quickly transferred to the tubi\s, 

 the delivery tuljes of Avhicli were in this case guarded Avith granular 

 calcium chloride. A tulie of dry potassium chlorate Avas heated Avith the 

 others, in order to judge the amount of expanding air forced over. 



