F. H. McCrudden and C. 8. Sargent 237 
Titration Error. 
Experi- Weight of : ae 
ment | approsimatsly | 9 APDpymately, | | Klused forsarh | _Dexiation fom 
gm ce. cc 
1 1.0462 5.75 5.495 0.032 
2 1.0492 5.79 5.519 0.008 
3 1.0482 5.78 5.514 0.013 
4 1.0482 9.79 5.524 0.003 
5 1.0489 5.80 5.530 0.003 
6 1.0486 5.79 5.522 0.005 
7 1.0473 5.82 DEOD 0.030 
Shy 1.0473 6.81 5.552 0.025 
MELE DCSE alt ire i. Ae oe oe 5.527 0.015 
The average deviation from the mean was 0.015 cc:; the extreme 
deviation only 0.03 cc. If we take 0.04 cc.—the weight of one 
drop of potassium iodide solution (average of six determinations) 
—as the titration error, the corresponding error in the (one- 
fifth as strong’) silver nitrate will be (one-fifth of 0.04 cc. =) 
0.008 cc. or 0.04 per cent (of 20 cc.).® 
The error in sodium (0.2 to 0.8 per cent) calculated from a 
0.04 per cent error in chlorine is reasonably close to that actually 
found by experiment (Table I and Chart 1). 
Only when five place logarithms are used is the computation 
error’ safely less than the analytical error? (Table II, Chart 2). 
5.8500 gm. of sodium chloride dissolved in water and made up to 1 
liter weighed 1,004.1 gm.; 1 gm. of the solution contained 0.0058261 gm. of 
sodium chloride. 7.4600 gm. potassium chloride dissolved in water and 
made up to 1 liter weighed 1,005.2 gm.; 1 gm. of the solution contained 
0.0074215 gm. of potassium chloride.° 
6 Nothing is gained by making the potassium iodide solution either 
stronger or weaker; with a stronger solution the error is still about one 
. drop—an increase in the absolute error. With a weaker solution, the 
absolute error increases proportionately; the proportional error remains 
the same. | ; 
7 The error coming from not preserving enough significant figures in the 
calculations. 
8 “Four place tables should not be employed upon work of an accur- 
acy exceeding one-half of one per cent,’’ Holman, Computation Rules and 
Logarithms, New York, 1913, p. Ixxvi. 
-®The pure chlorides were made from the purest obtainable com- 
mercial products by twice precipitating the salts from saturated aqueous 
solutions with hydrochloric acid gas. The quantities of these solutions 
used in all analyses were accurately weighed on the analytical balance. 
