156 R. N. Maxson — lodometric Determination of Gold. 



of the first series, the other solutions being of tlie original 

 strength, the total average error actually observed was consid- 

 erably lower, as would be expected, and amounted to + 0*00002 

 grm. of gold between extremes of +0*00006 grm. and —0-00002 

 grm. 



In a third series of ten experiments, in which the dilute gold 



N 

 solution (O'OSTl to 1 liter) was used, nearly iodine and 



Tsj- lUOO 



nearly thiosnlphate were employed. The average error 



amounted to less than 0*000004 grm. between extremes of 

 + 0*000020 grm. and —0*000029 grm., but in these determina- 

 tions the sensitiveness of the starch indicator to iodine becomes 

 an important factor, and it was necessarv to introduce a cor- 



"N . 

 rection of 0*1"^™^ for the amount of the iodine necessary 



to bring out the starch reaction in the volume of liquid used, 

 when tested in blank. 



In still another series of fourteen experiments, made upon 

 a gold solution (0*0104 to 200'"'') prepared by dissolving pure 

 gold foil in chlorine water and destroying the free chlorine by 

 ammonia, the average error amounted to +0*000002 grm. 

 between extremes of +000001 grm. and —0*000008 grm. 



The results of these experiments may be summarized as fol- 

 lows : 





Number 



of 

 deter- 

 mina- 

 tions. 

















Gold taken. 



Strength of solutions. 



Error. 

 Average. 



Extremes. 



Series I 



11 



8-71-s to 4-35'"g 



Iodine 



N 

 100 



Thiosnl- 

 phate 

 N 

 100 



Gold 

 0-871 



— 0-05™s 



j +0*03™e 



] -O-l'ng 



Series II 



20 



0-87™g to 0-087°^s 



N 

 100 



N 

 100 



0-0871 



4-0-02-!-' 



\ -t-O-OG^g 

 j _0-02"^g 



Series III 



10 



0-871"^sto0-087"'g; 



N 



fooo 



N 

 1000 



0-0871 



+ 0-004™^=' 



j +0-020'"g 

 ] -0-029-ff 



Series IV 



14 



0-520"^gto0-052'"s 



N 

 1000 



N 

 1000 



0-052 



+ 0-002'"? 



\ -fO-01™g 

 ] -0-008"? 



It is plain that the average experimental errors, due to all 

 causes, do not very much exceed the errors which might natu- 

 rally be expected to arise from errors of reading. In repeat- 

 ing the work of Gooch and Morley I have obtained results of 

 a reasonably similar order of accuracy. The process has, how- 

 ever, been recently criticized unfavorably by Eupp* on the 



* Ber. Dtsch. Chem. Ges., xxxv, 2011. 



