JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Precision of Methods Except for laboratory No. 2, determinations from the same laboratory were comparable in precision. Laboratories 1, 2, 4 and 5 obtained results which did not differ significantly in precision, but those from laboratory No. 13 were of lower precision. Apart from this laboratory the average precision per determination was between 4-5 cetrimide units. There was insufficient data to compare the precision of the two methods of analysis but there is an indication that the ptoluidine method is less precise than the methylene blue method. It will be necessary to base estimates of precision on 15-20 independent repetitions per analyst on each detergent sample in order to reach a definite conclusion about this, and to obtain at least 100 independent results from different analysts and labora- tories. The difference in relative accuracy between analysts in the same laboratory indicates a difference in analytical technique, such as in judging the end-point of titrations, which suitable training and collaboration might be expected to remove. The bias between the analysts in laboratory 5, affecting both methods of analysis must be due to some difference common to all determinations by these two operators. There is some evidence that the relative accuracy of the two methods of analysis varies between different laboratories. This is not conclusive, however, partly because the comparison is based on so few laboratories and partly because of the extent of variation between analysts in any given laboratory. Until this latter source of variation is eliminated (so that analysts in the same laboratory obtain consistent results) a definite con- clusion about variations between laboratories will not be possible. Table IV shows the average of determinations by each method for laboratories 1, 2, 4 and 5. Table IV Laboratory Laboratory Laboratory Laboratory Average 1 2 4 5 Method 1 ptoluidine 442 451.5 440 437 443 Method 2 Methylene Blue 458 454.5 477 465 461.5 Difference Method 1-Method 2 --16 --3 --37 --28 --18.5 Some approximate calculations of the size of investigation required to assess the significance of this variation show that data should be obtained

PRELIMINARY NOTES ON THE DETERMINATION OF ANIONIC SULPHATES 43 from 10-15 laboratories with two analysts per laboratory testing 2 or 3 samples of detergent in each case. REFERENCES (1) S. R. Epton Trans. Faraday Soc. 44 226 (1948) (2) D.C. Cullum Mfg. Chem. 33 352 (1962) (3) T. Bart, J. Ohrer and W. V. Stubbings J. Soc. Chem. Ind. London 67 2 (1948) (4) C. C. T. Chinnick and P. A. Lincoln •Proceedings, First World Congress Surface- Active Agents 1 Sect. 2 209 (1954) (Chambre Syndicale Tramagras, Paris) (5) T. U. Marron and J. Schifferli Ind. Eng. Chem. Anal. Ed. 18 49 (1946)