on the Analysis of the Carbonates. 349 



weight gives, by the usual procedure, the amount of the carbonate 

 in the known quantity of material used. 



Long experience with this instrument, especially as applied to 

 the alkaline and earthy carbonates, solid and in solution, has sat- 

 isfied us, that with the precautions above described, it yields 

 more uniform results than either of the methods commonly em- 

 ployed, while it possesses the important advantage of facility and 

 promptness in the manipulation. By comparing it with more 

 perfect arrangements, hereafter to be described, we have found 

 that with proper care, it enables us to ascertain the amount of 

 carbonate present, to within one tenth of a per cent., a degree of 

 accuracy, which, without the utmost precaution, is we believe, 

 rarely attained with Rose's apparatus, and which greatly exceeds 

 that of the operation with the graduated tube over mercury. 



The errors to which it is exposed, arise from two causes ; first, 

 the difficulty of removing the last traces of carbonic acid from 

 the air of the flask and pipette, by the pumping operation above 

 described, and secondly, the union of a portion of the carbonic 

 acid with the liquid in the flask. To these may perhaps be ad- 

 ded a slight endosmose through the gum-elastic bag, though of 

 this we have no certain evidence. Without therefore claiming 

 for it all the accuracy required in an instrument for refined re- 

 search, we offer it to the attention of practical chemists as a valu- 

 able help in the important class of chemical enquiries relating to 

 the composition of marls, calcareous soils and certain manufactur- 

 ed products, where despatch is of more importance than the high- 

 est degree of precision in the result. 



Of the sources of error above mentioned, that of the retention 

 of part of the carbonic acid by the liquid in the flask, is by far the 

 most important, and as will be shown hereafter, in the case of 

 Rose's process and its modifications by Fritche, and by Erdman 

 and Marchand, is of such magnitude, even when sulphuric acid 

 is employed, as greatly to impair the value of the results for the 

 purposes of nice investigation. 



Besides this source of inaccuracy, common to Rose's and our 

 own process, we have detected another peculiar to the former, 

 and which operates whenever hydrochloric acid is employed in 

 his apparatus. This is the evolution of carbonic acid from the 

 surface of the carbonate, caused by the action of the acid vapor, 

 during the time of the first weighing, and which occasions, as we 



