The effective temperature of a star may be defined, for the present purpose, as the temperature of a full radiator or "black body" for which the distribution of the energy in the spectrum most closely fits the observed distribution in the stellar It is assumed that if, as is usual, measurements are confined to those spectral regions which are free from absorption lines, the stellar distribution of energy is very nearly that of a black body; and this is probably the case. The principle of the method may be stated thus. If for two stars the ratio of the light reaching us be measured for two separate wave-lengths, and if the effective temperature of one of them, which is thus regarded as a standard star, be known, then the effective temperature of the other can be calculated. The problem is thus reduced to determining this ratio for two separate wave-lengths, or, which comes to the same thing, the difference of magnitude for the two wave-lengths in question. Actually it is not necessary to determine each of these differences, the difference of the differences is all that is required. The object of spectro-photometric measurement is thus to deter mine what is in reality a colour-index, determined with respect to two definite wave-lengths, say in the blue and the red regions of the spectrum. Then, if the effective temperature of the standard star be known, the effective temperatures of any others can be derived. The standard star need not be an actual star, but an
artificial source of light whose effective temperature can be de termined in the laboratory.
It is on these lines that research has been pursued at Pots dam, Edinburgh, Victoria, Greenwich and at other observatories. The work is still in its infancy and it is impossible to go into details here. With the exception of Potsdam, photographic methods are employed, and one method is to measure the densities of the silver deposited in the spectra with some form of micro photometer. It then remains to relate the measures of density to the absolute photometric scale just as in the case of ordinary photometry. A graded series of images in light of the requisite wave-lengths must be imposed on the plate and the measures of density thus calibrated In the case of the visual work carried out at Potsdam, the spectra were compared at each wave-length with a Nicol prism photometer in a manner analogous to the Harvard visual photometry. The results that have been so far obtained are not altogether accordant, and are subject to revision. The avail able observational data have been summarized by A. Brill (Astro nomische Nachrichten 5,539, 1928), who has given a table con necting spectral types and effective temperatures. According to this table, the mean effective temperatures of stars of type Bo is 20,800° Absolute. For type Ao the effective temperature is 1,300° A and for type Go (the solar type) it is 6,050° A for the "giants" and 6,240° A for the "dwarfs." For the very red stars of type Mo Brill gives 3,200° A and 3,420° A for the giants and dwarfs, respectively.
A very recent development is the determination, by photometric methods, of the integrated intensities of absorption lines, i.e., the ratio of the total amount of light absorbed in a line to the amount of light per unit of wave-length in the neighbouring continuous spectrum. Some very promising work on these lines has been carried out at Harvard and it is probable that it will be developed rapidly in the immediate future. (W. M. H. G.) PHOTON: see COMPTON EFFECT.