of error is as incumbent on an investigator of nature, as it is to enlarge the boundaries of science.

Chemists would do well to profit by the lessons to be derived from a study of the history of their science; for just as the animistic delusion of phlogiston dominated and retarded the progress of chemistry during the last century, so the notion of a periodicity in the atomic weights is the most remarkable illusion connected with the science of the present century, and equally detrimental to its future progress.

In conclusion, I would exhort all teachers of chemistry, whose aims are higher than to be mere phonographic exponents of other men's opinions, to examine for themselves the numerical relations of the atomic weights to which attention has been directed; as great responsibility rests upon those who continue the teaching of science in the Universities and Public Schools on foundations which a very limited exercise of the comparative faculty would prove to be false.

Ordinary Meeting, January 22nd, 1895.

Professor ARTHUR SCHUSTER, F.R.S., F.R.A.S.,
Vice-President, in the Chair.

The thanks of the members were voted to the donors of the books upon the table.

Mr. RUPERT SWINDELLS, M.Inst.C.E., exhibited six diagrams showing the continuous markings made by a thermograph and a barograph during the three recent storms of December 22nd and 29th, and January 12th. A discussion ensued, in which Professor OSBORNE REYNOLDS, Professor H. LAMB, Sir LEADER WILLIAMS, and Professor SCHUSTER, took part.

Professor ARTHUR SCHUSTER, F.R.S., read a paper "On a Comparison of the Thermometers used by the late Dr. Joule with the Standards of the Bureau International des Poids et Mesures." The Joule thermometers, and the apparatus by means of which the comparisons were made, were exhibited.

On a Comparison of the Thermometers used by the late Dr. Joule with the Standards of the Bureau International des Poids et Mesures. By Arthur Schuster, F.R.S.

(Received January 22nd, 1894.)

The mechanical equivalent of heat being defined as the quantity of work necessary to raise unit mass of water at a stated temperature through one degree, it is clear that its value must depend on what we adopt as the unit of our temperature scale. It is not sufficient to fix on the temperature of freezing or boiling water, and to call them o° and 100° respectively, for this leaves it still doubtful what we should take to be 50° or 60°. It has been usual for a long time to take the expansion of mercury in a glass vessel as our guide, and to fix the temperature intermediate between 0 and 100° by means of the mercury thermometer. For many purposes this is sufficiently accurate, but the irregular behaviour of such thermometers has led to some difficulties which have only been recently surmounted. The nature of the glass having a considerable influence on its coefficient of expansion, it has become necessary carefully to compare thermometers made of different kinds of glass, and those who have followed the recent development of this subject will know of the great progress which has been made in this direction by the Technische Reichs Anstalt and the Bureau International des Poids et Mesures.

Since the late Dr. Joule concluded his classical researches, several accurate re-determinations of the equivalent of heat have been made, and in order to compare the results obtained by different observers it seemed to me to be of great interest to find the scale value of Joule's thermometers in terms of

some fixed or easily reproducible scale. The request which I made to Mr. B. A. Joule to allow me the use for a short time of his late father's thermometers was met at once with a most ready compliance, for which I have to offer my best thanks.

Thermometers made of hard French glass, having a definite chemical composition, have been investigated with great detail by M. Guillaume, of the Bureau International, and have been compared directly with the air thermometer by M. Chappuis.

I have thought it advisable, therefore, to compare the scale value of Joule's instrument in terms of that of a thermometer made of French standard glass.

The two thermometers which I had at my disposal were then called A and D in Joule's published papers; they were made and calibrated by Dancer in 1844. The thermometer D contains both the freezing and boiling point, while A only reaches up to about 30°C.

All observations on the equivalent were made with A, the scale value of which was determined by a comparison with the standard D. One division of A was found in mercury by Dr. Joule to correspond to o°077214 F., which is equal to o°042897 C.

To determine the relation between these thermometers and those now in use, it seemed sufficient to confine the investigation to the thermometer A. The thermometers having been calibrated so that the divisions are intended to represent equal volumes of the capillary tube, the distances of the divisions may serve as a test of the equality of the bore at different places. Near the freezing point the length of 50 divisions is 43 cms.; this length increases gradually until at the other end of the scale 50 divisions occupy a length of 5.3 cms.; the tube is therefore conical, the diameter at the top being about ten per cent smaller than at the lower end. A careful examination of successive lines on the stem