VOL. LXXIV.] PHILOSOPHICAL TRANSACTIONS. 50Q 



however conclude, without acknowledging my obligations to Dr. Priestley, who 

 has given me every information and assistance in his power, in the course of my 

 inquiries, with that candour and liberality of sentiment which distinguish his 

 character. 



XXVI. Sequel to the Thoughts on the Constituent Parts of Water and Dehhlo- 



gisticated Air. In a subsequent Letter from Mr. Watt to Mr. De Luc, F. R. S. 



p. 354. 



On re-considering the subject of my letter of the 26th of Nov. last, I think 

 it necessary to resume the subject, in order to mention some necessary cautions 

 to those who may chuse to repeat the experiments mentioned there, and to point 

 out some circumstances that may cause variations in the results. In experiments 

 where the dephlogisticated air is to be distilled from common or cubic nitre, 

 these salts should be purified as perfectly as possible, both from other salts and 

 from phlogistic matter of any kind ; otherwise they will produce some nitrous 

 air, or yellow fumes, which will lessen the quantity, and perhaps debase the 

 quality of the dephlogisticated air. If the nitre is perfectly pure, no vellow 

 fumes are perceptible, until the alkaline part begins to act on the glass of the 

 retort, and even then they are very slightly yellow. When earthen retorts are 

 used, and a large quantity of air is drawn from the nitre, it acts very much on 

 the retort, dissolves a great part of it, and becomes very alkaline, retaining only 

 a small part of its acid, at least only a small part which can be made appear in 

 any of the known forms of that acid ; and unless retorts can be obtained of a 

 true apyrous and compact porcelain, I should prefer glass retorts, properly coated 

 for making experiments for the present purpose. 



In some of my experiments the nitre was left in the retort placed in a furnace 

 so that it took an hour or more to cool. In these cases there was always a de- 

 ficiency of the aciil part, which seemed, from some appearances on the coating, 

 either to have penetrated the hot and soft glass, by passing from particle to par- 

 ticle, or to have escaped by small cracks which happened in the retort during the 

 cooling. There was the least deficience of the acid when the distillation was 

 performed as quickly as was practicable, and the retort was removed from the fire 

 immediately after the operation was finished. In order to shorten the duration 

 of the experiment, and consequently to lessen the action of the nitre on the re- 

 tort, it is advisable not to distil above 50 oz. measures of dephlogisticated air 

 from an ounce of nitre. The experiment has succeeded best when the retort was 

 placed in a charcoal fire in a chafing-dish or open furnace ; because it is easy in 

 that case to stop the operation, and to withdraw the retort at the proper period. 



When the dephlogisticated air is distilled from the nitre of mercury, the solu- 

 tion should be performed in the retort itself, and the nitrous air produced by the 



vol. xv. 4 D 



