RECOLLECTIONS AND REFLECTIONS
centred on the accurate measurement of electrical resistance. Much work had been done in this subject, and methods evolved which gave results of great accuracy. The instrument required—the galvanometer—had been made so reliable that the use of it did not require a long apprenticeship. The resistance of a platinum wire depends upon its temperature, so that if the resistance of the wire at different temperatures is known you can, by measuring its resistance, determine its temperature. The wire acts like a thermometer, and since platinum only melts at a very high temperature, it can be used for measuring temperatures at which a mercury thermometer would be useless. Siemens had actually constructed a thermometer on this principle, but this was found to have grave defects which made accurate determinations of temperature impossible. The simplicity and convenience of using a piece of wire as a thermometer was so great that it seemed to me very desirable to make experiments to sec if the failure of Siemens' instrument was inherent to the use of platinum as a measure of temperature, and not to a defect in the design of the instrument. Callcndar took up this problem with great enthusiasm and showed that, if precautions are taken to keep the wire free from strain and contamination from vapours, it makes a thoroughly reliable and very convenient thermometer. This discovery, which put thermometry on an entirely new basis, increasing not only its accuracy at ordinary temperatures, but also extending this accuracy to temperatures far higher and far lower than those at which hitherto any measurements at all had been possible, was made with less than eight months' work. I had very little to do with it beyond seeing how it was going on from day to day, and encouraging him when he was disheartened by the set-backs which occur in all 132