It is not the province of a Compiler to accept or reject data according to their aspect of verisimilitude or otherwise. For such complex substances as oils, fats and waxes - many of them inevitably mixed with other bodies from the very nature of their commercial production - seemingly discrepant data cannot be rejected off-hand. Indeed, in many cases, when taken in conjunction with the information supplied in the text of the published source, an abnormal figure may be a useful index to the origin of the fatty substance or to the process by which it was obtained.

The Compiler, who attaches the utmost importance to the accurate determination of the temperature coefficient - far greater importance than it seems to have received in the past - has made a point of stating its value wherever an author has either given it expressly or where two or more readings for different temperatures have enabled the Compiler to calculate it. These latter calculated values are marked by an asterisk appended to the reference numbers in the last column.

From the recent paper of Wright and that of Joseph, to which the Editor refers in the Introduction, and also from the majority of values given in these tables, 0.00037 may be taken as a probable mean value at temperatures not too far removed from 25° and 40° C. There are a number of δn/δt values which differ widely from the above mean. These temperature coefficients would seem to be at least a check upon the refractive constants given for different temperatures, but in some cases they clearly invite caution. For instance, when the temperature coefficient, as derived from two nD values, showed a doubtful δn/δt value, 0.00126 say, the Compiler did not hesitate to reject it.

The advantage of knowing the temperature coefficient between 15° and 40° C. is that it is much easier to take a refractometer reading at a temperature which has been obtaining steadily for some time, and to reduce the observation to a standard temperature, than it is to establish the required temperature.

The Compiler expresses the hope that the practical utility of these tables may be enhanced by a series of critically accurate determinations of temperature coefficients, so that it may be practicable in future to reduce all data to one or two standard temperatures.

The Compiler takes this opportunity to express his grateful thanks to the Editor for the untiring help which he has given him in what was necessarily a laborious task.

To Messrs Adam Hilger, Limited, the Compiler wishes to express his sincere thanks for having enabled him to perforin his task with all the freedom which is needed for scientific research.

R. K.