OF ARTS AND SCIENCES. 239 



which had given the substance melting at 185", turned out to be too 

 wasteful a process for working up large amounts ; we were not sur- 

 prised, therefore, that our analysis of the silver salt gave 62.82 per cent 

 of silver, instead of the 61-. 69 per cent calculated for argentic tarti o- 

 nate. This result, however, bad as it is, comes nearer to that for argen- 

 tic tartronate than to that for any other substance which could well be 

 formed, and therefore gives us the desired confirmation.* The small 

 amount of tartronic acid obtained, less than 0.5 gr. from about 10 grs. 

 of the oil, may be accounted for on the supposition that the larger 

 part of the tartronic acid was decomposed by the strong hydrochloric 

 acid used in the saponification of the oil, as Conrad and BischofFf have 

 observed that tartronic acid is decomposed very easily by hydrochloric 

 acid, and we also were unable to obtain a trace of the acid from the 

 filtrate after argentic chloride had been precipitated from the silver 

 salt by a slight excess of hydrochloric acid. In fact, we should not 

 have thought it worth while to try to obtain tartronic acid by the ac- 

 tion of hydrochloric acid on our oil, if it had not been for the fact 

 that tartronic ester has been made from calcic tartronate by the action 

 of hydrochloric acid gas and alcohol,^: which would seem to show that 

 the tartronic acid is more stable in presence of strong than of dilute 

 hydrochloric acid, improbable as this sounds. This view is in harmony 

 with our observations detailed above, and Conrad and BischofF state 

 that even very dilute acid decomposes it readily ; but even granting 

 this, a large proportion of the tartronic acid must have been decom- 

 posed during the treatment with hydrochloric acid, and this seems to 

 us sufficient to make it more than probable that the tartronic acid is 

 the product of the principal reaction, and not formed in a secondary 

 one taking place only to a limited extent. If this is admitted, the fol- 

 lowing reactions would give the most probable explanation of what 

 takes place. 



I. C6HBr3(NO,)2 + NaCH(COOC2H,)2 = 



C6HBr,(N02)2CH(COOC2H5)2 + NaBr. 

 II. C6HBr2(N02\,CH(COOC2H,)2 + NaCH(C00C2H,), = 



C6HBr2(NO,),CNa(COOC2H5), + CHo(Co6c2H,)2. 



* No attempt was made to obtain a better analysis of the tartronic acid, 

 because the work upon the corresponding trinitro compound described in an- 

 other paper furnished a complete confirmation of the theory of the reactions 

 given here. 



t Ann. Chem., ccix. 223. 



t Freund, Ber. d. ch. G., xvii. 786. Pinner, Ibid., xviii. 757. 



