Our methods of enlargement have been considered, and some of them tried, with the object of removing the irregularities of the original spectra without introducing new defects.  For instance, the sensitive plate may be moved during the enlargement in the direction of the spectral lines; a slit parallel to the lines may be used as the source of light, and the original negative separated by a small interval from the plate used for the copy; or two cylindrical lenses may be used, with their axes perpendicular to each other.  In some of these ways the lines due to dust might either be avoided or so much reduced in length as not to resemble the true lines of the spectrum.

The 15 inch refractor is now being used with a modification of the apparatus employed by Dr. Draper in his first experiments—­a slit spectroscope from which the slit has been removed.  A concave lens has been substituted for the collimator and slit, and besides other advantages, a great saving in length is secured by this change.  It is proposed to apply this method to the 28 inch reflector, thus utilizing its great power of gathering light.

[A description of an accompanying plate here follows, which is omitted, as the plate cannot be easily reproduced for ordinary press printing.]

The results to be derived from the large number of photographs already obtained can only be stated after a long series of measurements and a careful reduction and discussion of them.  An inspection of the plates, however, shows some points of interest.  A photograph of a Cygni, taken November, 26, 1886, shows that the H line is double, its two components having a difference in wave length of about one ten-millionth of a millimeter.  A photograph of o Ceti shows that the lines G and h are bright, as are also four of the ultra-violet lines characteristic of spectra of the first type.  The H and K lines in this spectrum are dark, showing that they probably do not belong to that series of lines.  The star near [chi]’ Orionis, discovered by Gore, in December, 1885, gives a similar spectrum, which affords additional evidence that it is a variable of the same class as o Ceti.  Spectra of Sirius show a large number of faint lines besides the well-known broad lines.

The dispersion employed in any normal map of the spectrum may be expressed by its scale, that is, by the ratio of the wave length as represented to the actual wave length.  It will be more convenient to divide these ratios by one million, to avoid the large numbers otherwise involved.  If one millionth of a millimeter is taken as the unit of wave length, the length of this unit on the map in millimeters will give the same measure of the dispersion as that just described.  When the map is not normal, the dispersion of course varies in different parts.  It increases rapidly toward the violet end when the spectrum is formed by a prism.  Accordingly, in this case the dispersion given will be that of the point whose wave length is 400.