December, 1903.] 



KNOWLEDGE 



273 



This structure is better seen in the enlarged view of one 

 of the flutings given in Fig. 12. Here, the flutings fade 

 to the violet, but in some cases the degradation is towards 

 the red. Laboratory experience teaches tliat flutings only 

 appear in the spectra of sources of liglit which are at a 

 relatively low temperature as compared with that necessary 

 for the production of the line spectra of the same sub- 

 stances ; their presence is, therefore, a valuable clue to the 

 temperature conditions which prevail in the atmospheres 

 of the stars which exhibit them. Many flutings originate 

 in compounds, those at A. \ 3883, 4216, and 4009 in the 

 arc sj^ectrum of carljon, for example, being almost certainly 

 due to cyanogen, and the disappearance of such flutings 

 at increased temperature is sufficiently explained by the 

 breaking up of the compound into its constituent elements. 

 The replacement by lines of flutings of elementary sub- 

 stances, such as those of carbon at 4737, 5165, and 5635, 

 may be supposed, as suggested by Lockyer, to be brought 

 about by the separation of complex molecular groupings 

 into simpler ones. 



Extensive observations of the spectra of third and fourth 

 type stars (or Antarian and Piscian stars, in Lockyer's 

 nomenclature) have been made by Duuer, Vogel, Huggins, 

 Espin, and others, but visual observations, though of 

 intense interest to the observers, are now quite superseded 



On account of their feeble luminosity, none of them 

 being brighter than 5th magnitude, the spectra of the 

 fourth type stars can only be photographed with great 



JiU.i. 



Fm. 14..— Spectra of (I.) Sun, (II.) /^ Geminorum, (III.) 132 Sehjel- 

 lerup. From photographs taken at the Yerkes Observatory. 



difiiculty. McCleau, however, succeeded in photographing 

 I two of the brightest specimens, and discovered that besides 



Fig. 13. — Spectra of o Ceti and a Heroilis, from photographs taken at Stonyhurst College. Tlie wave-lengths of the 

 numbered lines and bands are:-(l) 4227, (2) U20, (3) 4581, (4) 4757, (5) 4951, (6) 5165, (7) 5447, (S) 5597, (9) 5756. 



by photographs in aU investigations involving accurate 

 measurements. Many successful photographs of third 

 type spectra have been secured at various observatories, 

 and wc are fortunate in having the kind jierinission of 

 Father Sidgreaves to reproduce two lieautiful exami)les 

 from the Stony hurst collection, illustrating the spectrum 

 in the yellow, green, and blue, a Herculis may be 

 considered a normal third tyj)e, while the photograph 

 of Mira admirably illustrates the introduction of bright 

 hvdrogon lines which is, in some way, associated with the 

 variability of that star, and is, m fact, characteristic of 

 the numerous variables having spectra of the third type. 



the flutings the spectra contain numerous lines which 

 gi-eatly resemble those of a Tauri. Unexpectedly fine 

 photographs, exliibiling a wealth of detail, have since been 

 obtained at the Yerkes Observatory, where the great light- 

 gathering power of the 40-inch refractor has allowed the 

 use of considerable dispersion. Through the kindness of 

 Prof. Hale, we are able to reproduce the interesting series 

 of photographs in Fig. 14, comparing the yellow and green 

 parts of the spectrum of a third and a fourth type star 

 with that of the sun. 



The number of third type stars already known is esti- 

 mated by Miss Gierke at about 2000, and that of the 



