they must necessarily be at all times symmetrically placed; so that if 1 be fixed and motion be communicated to 4, then 2 and 3 will move, each in its proper degree, and all four will preserve their proper relative positions. In Browning's spectroscope there are six such prisms, and the light emerging from the sixth passes just clear of 1.

I have suggested a modification, by which only a corner of the first prism is fixed, as at fig. 30, and this

prism is automatically adjusted like the rest,-a fixed slots s' square to the path of the incident light guiding the motion of this prism according to the required law. In like manner I have suggested that the viewing telescope should be guided on the same principle of automatic motion.*

This arrangement must be regarded as only one out of several modifications of which Mr. Browning's ingenious plan admits. It is doubtless this modification which alone satisfies the conditions aimed at; and this circumstance renders it all the more evident that it should be regarded as implicitly contained in Mr. Browning's plan. Nothing is easier than to devise modifications, and even (as in this instance) improvements on the plans of others. But no separate merit can fairly be claimed under such circumstances. In this special instance, for example, I should not have turned my thoughts towards the problem of securing the conditions of minimum deviation for all rays, but for the interest I took in Mr. Browning's plan. So soon as he had read the paper describing his plan, before the Royal Astronomical

So far as clearness of definition and the satisfactory study of the whole length of the spectrum are concerned, this arrangement leaves nothing to be desired. But for a yet greater increase of the spectrum's length more is needed. We have reached the limits of dispersion which one circular battery can give; but it would be desirable to obtain a yet greater dispersion.

One way in which this can be done is by the use of what are called direct-vision prisms. In these, two flintglass prisms (FF) are combined, as shown in fig. 31, with three crown-glass prisms. The prisms c cause

deviation and dispersion in one direction; the prisms F cause equal deviation and more dispersion in the contrary direction. Hence results a balance of dispersion without deviation; and if we add such a prism to any battery of prisms, we get all the advantage of the dispersion without increasing the deviation which had been our difficulty.

Society, I felt that his method involved the complete solution of the difficulty. He is now constructing an automatic spectroscope, in which the modifications I have suggested have been introduced; and as complete success depends only on the exactness with which the mechanical and optical relations are fulfilled, there can be no doubt of the result, for rigid accuracy is a characteristic feature of all Mr. Browning's work. Let it be understood clearly that if I have any merit in the matter at all, it consists in pointing out two unnoticed good qualities of Mr. Browning's own plan. Modified or not, the plan is altogether his.

Yet there is one important disadvantage in directvision prisms, more especially when they are employed in researches requiring very neat and exact definition: it is, of course, wholly impossible to employ any method for securing minimum deviation.

A plan by which the dispersion in a battery of prisms may be doubled, falls next to be considered. Supposing A and B (fig. 32) to be two of the ordinary triangular prisms and c a right-angled prism half the size of the others, then the rays which fall on CD are reflected back again through the battery. I need not

here describe the contrivance by which the emergent pencil can be viewed in such a case by a telescope in which there is a totally reflecting prism sending the light out at right angles to the axis of the tube. According to another arrangement, in place of such a prism as c, one is employed which by two total reflections raises the rays to a higher level, so that (the prisms of the battery being twice as high as usual) the light returns by a separate course.

I have devised a plan by which a much greater dispersion than has ever yet been gained may be combined with a perfectly true automatic adjustment for

all parts of the spectrum. In this plan Mr. Browning's automatic method is extended to a second battery, while the plan for returning the rays, illustrated in fig. 33, is employed in such sort as to double the dispersive power of the double battery. Fig. 33 shows the arrangement of the two batteries. A B is the light incident on the first battery, and the course of the light can be traced by the triple set of lines through the

double set of prisms, the dotted return lines showing the course of the return rays to emergence at cc'. It will be seen that the large intermediate prism DE may be regarded as belonging to both batteries. There is no loss of light in passing from one battery to the other, since the reflection at D D' is total. The close double lines show the direction of the slotted bars, E F being a long slotted bar kept square to the rays leaving the face D'E by means of the equality of G F and G E.

(It pivots round E and slots at F.)* The figure shows the arrangement of a battery having an effective dispersing power equal to that of nineteen equilateral prisms of heavy flint-glass.†

It must be remembered that what the spectroscope really does is to give a range of pictures of whatever

FIG. 34.

S

R

luminous object or part of an object would be visible through the slit if the spectroscope were removed.

It is not improbable that before these lines are read an instrument on this plan will have been constructed by Mr. Browning, to whom I have submitted the proposal. I cannot but think that if the mechanical and optical difficulties which the plan involves can be overcome (and if he cannot overcome them I know not who can), such an instrument will prove of considerable service in researches into solar physics. It will be seen that the plan is only available where very strong light is to be analysed. But in dealing with the Sun this consideration introduces no new difficulty.

G

+ The double battery is represented in about the position corresponding to minimum deviation for the line a in the indigo; and as the half prism can be made to pass over the slot at F, the automatic motion can be carried on till the visible extremity of the spectrum towards the violet is reached. Of course, in examining the red rays, the batteries open out, and there is no limit (except the length of the slots) to motion in that direction, so that the red extremity of the spectrum can be overpassed if need be. Motion is communicated to the intermediate prism D D', by which means the range of the automatic action is divided, and greater truth of working secured.