It may be pointed out here that no single pyrometer is suited to every purpose, and the choice of an instrument must be decided by the nature of the work in hand. A pyrometer requiring skilled attention should not be entrusted to an untrained man; and it may be taken for granted that to obtain the most useful results intelligent supervision is necessary. In the ensuing pages the advantages and drawbacks of each type will be considered; but in all cases it is desirable, before making any large outlay on pyrometers, to obtain a competent and impartial opinion as to the kind best suited to the processes to be controlled. Catalogue descriptions are not always trustworthy, and instances are not wanting in which a large sum has been expended on instruments which, owing to wrong choice, have proved practically useless. An instrument suited to laboratory measurements is often a failure in the workshop, and all possibilities of this kind should be considered before deciding upon the type of pyrometer to be used. CHAPTER II STANDARDS OF TEMPERATURE =The Absolute or Thermodynamic Scale of Temperatures.=—All practical instruments for measuring temperatures are based on some progressive physical change on the part of a substance or substances. In a mercury thermometer, the alteration in the volume of the liquid is used as a measure of hotness; and similarly the change in volume or pressure on the part of a gas, or the variation in resistance to electricity shown by a metal, and many other physical changes, may be employed for this purpose. In connection with the measurement of high temperatures, many different physical principles are relied upon in the various instruments in use, and it is of the greatest importance that all should read alike under the same conditions. This result would not be attained if each instrument were judged by its own performances. In the case of a mercury thermometer, for example, we may indicate the amount of expansion between the temperatures of ice and steam at 76 centimetres pressure, representing 100° Centigrade, by _a_; and then assume that an expansion of 2_a_ will signify a temperature of 200°, and so on in proportion. Similarly, we may find the increase in resistance manifested by platinum between the same two fixed points, and indicate it by _r_, and then assume that an increase of 2_r_ will correspond to 200°. If now we compare the two instruments, we find that they do not agree, for on placing both in a space in which the platinum instrument registered 200°, the mercury thermometer would show 203°. A similar, or even greater, discrepancy would be observed if other physical changes were relied upon to furnish temperature scales on these lines, and it is therefore highly desirable that a standard independent of any physical property of matter should be used. Such a standard is to be found in the thermodynamic scale of temperatures, originally suggested by Lord Kelvin. This scale is based upon the