plane grating, with ruled surface 5×3 inches [127 x 89 mm.] with 14,438 lines to the inch, is used to produce the spectrum. In order to allow of the use of spectra of high order without the troubles introduced by superposed spectra, a prism of small angle 5° and of circular aperture 4 inches [101 mm.] has been inserted, with its refracting edge perpendicular to the rulings of the grating between the grating and the object-glass. Both object-glass and prism are the work of Messrs Cooke & Sons of York, and are of excellent performance. Provision is made for tilting the spindle and frame, which hold the grating, about an axis parallel to the refracting edge of the prism; in this way any part of any spectrum is brought to the centre of the photographic plate, and the dispersion is such that the ordinarily superposed spectrum is thrown above or below. Then, on turning the grating about its axis of rotation, the succession of appearances in the field of view is such that the blue end of the spectrum is seen at the bottom of the field of view; as we turn towards the green, the spectrum mounts to the middle of the field; and on turning further, the red is to be seen at the top of the field. By the tilting of the spindle

on which the grating turns, any part of the spectrum can be brought to the middle of the field.

The instrument, which is of very rigid construction and weighs about 6 hundredweight (300 kilogrammes), has been mounted horizontally on roller bearings, which allow of easy rotation of the whole about a horizontal axis, so that the slit can be turned into any position angle. A position circle is provided, divided to half degrees. The axis on which the instrument turns coincides with the line joining the middle point of the slit to the middle point of the object-glass. In this way a horizontal pencil of rays incident from the proper direction on the centre of the slit always passes through the centre of the collimator-camera lens, whatever may be the position angle of the slit. The true optic axis passes through the focal plane at a point midway between the middle point of the slit and its image; this point thus describes a small circle round the middle of the slit as the whole instrument is rotated on its roller bearings. The middle point of the photographic plate similarly describes a concentric circle of twice the radius.

Black opaque diaphragms of the proper size have been inserted in the proper places between the slit and the object-glass, in order to prevent the light reflected from the surface of the lenses from falling on the spectrum on the photographic plate.

A convenient device has been added to limit the length of the slit used, and to allow of the juxtaposition of spectra in successive exposures on one and the same photographic plate; thus a spectrum of the east limb of the Sun can be photographed between two spectra of the west limb, or two spectra of the east limb can be alternated with two spectra of the west limb.

In the preliminary work with the instrument great care had to be exercised in overcoming the erratic performance of the grating; and in order to secure the defining power required for this work it was found necessary to limit the area of the ruled surface by means of diaphragms carefully placed. When all the ruled surface except a strip 5 inches wide and about 1 inches high was blocked out, it was found that excellent definition could be attained with full resolving power; and the exposures needed for photographing the cyanogen band at 3883 in the spectrum of the third order seldom exceeded a minute, and in a bright sky an exposure of 30 seconds was ample.

The photographs obtained with the instrument exhibit the spectrum with very open scale, and many of the close double lines even in the cyanogen baud are well seen. The linear dispersion of the negatives is such that we have about 129 tenthmetres per millimetre. And it should be remembered that the total exposure of a plate in the instrument was completed generally in less than 3 minutes.

Examination of the photographs obtained disclosed peculiar features. It was at once evident that the spectrum of the east limb was nearly always less well defined than the spectrum of the west limb over a considerable region of the spectrum near the cyanogen bands. The photographs dealt with may be divided into two classes: firstly, about 35 "adjustment photographs" taken for finding the best adjustment of the instrument and for testing the effects of some exaggerated maladjustments; and secondly, about 10 "rotation photographs," each exhibiting two spectra of the east limb alternating with two spectra of the west limb, all taken with approved adjustment of the instrument.

A great deal of care and time have been devoted to the critical study of the evidence; and though at first I was led to suspect that, in spite of great care in adjusting the instrument, and in making sure that the optical beam was arranged exactly similarly both for east limb and for west limb, some unexpected error had crept in, I am now convinced that the right interpretation of the peculiar effects lies in accepting them as real, and not instrumental.

The spectrum of the east limb was in nearly every case less clearly defined than that of the west limb. Now, in a malfocussed photograph, one can, from the appearance presented by a single isolated line, predict with considerable certainty what the appearance of various groups of lines will be. But the appearances presented by the spectra of the east limb do not agree with prediction based on study of isolated lines or edges. In particular, the cyanogen band itself at the east limb shows in general a double edge, as if

two similar bands are superposed, with slight relative displacements which vary from day to day. The cyanogen band at the west limb shows in general a single slightly intensified edge, as if there might be two bands superposed, with such slight displacement as not to be measurable. There are many other minute differences between corresponding lines in the spectra of the east and west limbs. Three of the "rotation photographs ing the best definition: namely

L. 32

1907 Sept. 15, 10.23 a.m.

were picked out as show

G.M.T (civil).

These have been measured by my assistant, W. H. Manning, under my direction. Six iron lines, two chromium lines, and five cyanogen lines were chosen for measurement between wave-lengths 3871 and 3898. Each plate was measured twice in reversed positions on the Zeiss Comparator. The values of the displacements (OR), East-West, are given for each line in the following table; the values are expressed in millimetres. Four of the five cyanogen lines are really close double pairs, which were bisected as single lines. They are indicated in the table.

It thus appears that the cyanogen lines give displacements (and consequently also a solar rotation-period) sensibly the same as those given by iron lines of solar origin. These displacements of lines relate to spectra taken from solar light gathered at the east and west ends of the solar equator, within about 5° of solar latitude from the equator, at points situated within the limb by about of the radius of the Sun's apparent disc. I do not put them forward as satisfactory determinations of the equatorial velocity of the Sun. To deduce the velocities corresponding to the displacements, we must use a factor, L/λ dλ/dR, which works out to be sensibly 100. For the velocity per tenthmetre (L/A) is

and

whilst the linear dispersion is such that dλ/dR is equal to

Thus the factor ranges from 1006 to 998, and for our present purposes we may take its value as 100 over the range of spectrum measured.

Hence the displacements expressed as velocities are

for the 6 iron lines, 3.27 km/sec

The corresponding equatorial velocities are got by halving these numbers. It is seen that the uncorrected numbers are a little lower than Dunér's and Halm's.

The main point, however, is that for these three determinations cyanogen is shown to be in the solar atmosphere in the same sense that iron and chromium are.

The wave-lengths of the cyanogen lines used in this determination have been also deduced with reference to three iron lines as standards. They are found to differ from Rowland's values by less than one hundreth of a tenth metre, as will be seen from the accompanying table.

The apparent duplication of the edge or head of the CN band is, as has already been stated, more marked at the east limb than at the west. It is much more difficult to measure the displacement of edges than the displacement of lines in the band. Nevertheless I have made some measurements of the displacement; and if these are interpreted as indicating a Doppler effect, the results may be summarised as follows, the velocity of the east limb always being towards Earth.

This progressive change of the duplication of the edge is seen to go on for 9 days, and so can hardly be connected with ordinary disturbances of a local nature on the Sun's surface. The cyanogen involved cannot be far from the Sun, for otherwise it would be detected in the spectra of both limbs. It is seen that nothing of the order of cometary velocities is as yet detected.

I have spent some time in considering the possible influence of Daniel's comet which reached perihelion on September 4, with a velocity of 58 km/sec at a distance 0.512; the Earth passed closest to the path of the comet on September 12, as was pointed out by Kritzinger (Ast. Nach., 4198). The comet itself had passed the point on July 27. But in the solar spectrum no sign has been detected that can fairly be attributed to the presence of the comet near the Sun.

Much more evidence is required before the case can be completely elucidated, and this note must be regarded as of a preliminary nature.