400 Imperial Institute of France. 



worthy of trial, and accordingly cuttings of full bleached 

 linen were steeped in soUitions of the neutral salt, of different 

 degrees of strength, some of them nearly saturnted ; and 

 after conlinuing the process with fresh solutions for four 

 times, each twenty-four hours, in every experiment, and 

 at the end con)parmg the linen with some of the same 

 which had not been steeped, it was found perfectly sound, 

 and not perceptibly reduced in the texture. It was there- 

 fore stated as probable, that if accidents ever occur in the 

 process it must be nwmg to disengaged nmriatic acid, or 

 to negligence \\\ washing the cloth atler the operation of 

 sle»'ping. 



The assertion of Mr. Davy that the oxyinuriate of mag- 

 nesia has superseded the use of that of lime, in Ireland, 

 was contradicted. Not a single bleacher in the country 

 uses it; for, if eligible even, it is not within his reach, mag- 

 nesia being <2s. or 3s. per pound. It uppears from the arti- 

 cle jBlt:acuing, iu the Edinbvirgh Cvclopacdia, that the oxy- 

 munate of magnesia has been employed by the calico 

 printers ol Scotland, in the process of clearing, for some 

 time back. Air. Davy is not mentioned as the proposer of 

 it in that article. 



IMPERIAL XXSTITUTE OF FRANCE. 



[Continued from p. 244.] 



The author adds the following rcflecfinns which are com- 

 pletely independent of the theory of probabilities, on which 

 he founded the preceding reasoning. 



The system of elements which most reduces the errors 

 will certainly be the iriost probable, if all the observations 

 be equally accurate: but if there are two systems of ele- 

 ments, iMie of which represents, in the best manner, a cer- 

 tain number of observations, and the ')ther of which best 

 agrees with other observations, then we fall again into per- 

 plexity anil uncertainly, and we may propose innumerable 

 svstenis lor lessening the errors: we may instead of the 

 small squares propose small equal powers of ativ given or- 

 der, but the squares arc always the most simple — the other 

 powers would lead us into endless calculations. If the ex- 

 j>onent pair of powers is infinite, we recur to the method 

 which demands that the extreme errors shall be minima. 



He finds that the principle of Bo«covich returns to the 

 method in vvli ch it might be proposed to satisfy rigorously 

 a number of equations equal to that of the unknown quan- 

 tities, and in whicli we should onW consider all the rest as 

 so many proofs to enable us to judge of the precision 



which 



