insects are exceedingly difficult to make out; but I think I have succeeded in working out the retinal structures of Eristalis and Agrion in considerable detail. In the other insects in which I have examined the eye, the knowledge which I have been able to obtain of this portion of the nervous system, must be considered at present as fragmentary. In Eristalis there are from without inwards:-1. A double layer of large stellate ganglion cells; 2. A layer of small round nucleated cells; 3. The facelloid layer already referred to; and 4. A layer of stellate ganglion cells.

These parts are connected with a deep ganglion, which consists of several layers of fusiform cells by a decussating optic nerve, the fibres of which cross each other from above downwards, and from without inwards. The deep ganglion is connected by a distinct peduncle with the supra-cesophageal ganglion. All the structures of the ganglionic retina are supported by a fine neuroglia, which extends from a thick outer to a fine inner limiting membrane.

In Agrion the ganglionic retina differs from that of Eristalis in the absence of a facelloid layer, which is replaced by a triple layer of prismatic cells the investigation of the nerve structures of this insect is rendered very difficult by the presence of a large quantity of dark pigment in the stellate connective cells of the neuroglia.

In Vanessa the facelloid layer of the retina is also absent, but in its place there are numerous layers of fusiform cells. In noctuid moths, or at least in some species, the nervous structures are obscured by the large quantity of deep black pigment which they contain. In the semi-compound, or, as I have termed it, the microrhabdic eye of Tipula, the nervous retina consists of, (1,) a layer of stellate ganglion cells; (2,) of several layers of round cells; and (3,) of several layers of fusiform cells. The greatest simplicity exists in the eye of Formica, in which all the structures of the nervous retina are absent except numerous layers of small round cells. I have not at present been able to make out any decussation of the nerve fibres connecting the deeper parts with the ganglionic retina in the insects with microrhabdic eyes, but the investigation is very difficult, owing to the great change of the plane in which the nerve tracts lie. I do not think, however, that any decussation exists, or I think I should have found indications of it.

The extent and curvature of the cornea and the size and curvature of the facets afford the most important indications as to the manner in which vision is accomplished. In the true compound eye, I think the structure indicates that J. Müller's theory of vision is the most probable; this is also Dr. Grenacher's view, and it is supported, as I shall now endeavour to show, by the curvature of the cornea and the size of the corneal facets in different insects, as well as in different parts of the same eye.

The semi-compound eye introduces no new difficulty in this theory, it is probable, I think, that more than a single luminous impression is received by the elements which are situated behind each facet, and that these correspond with portions of the field of vision which are remote from each other, the central rod cell of one facet corresponding to one of the peripheral rod cells of some other facet; the extreme complexity of the connexions between the cells of the ganglionic retina renders this view not improbable.

In order to determine the effect of the long fine highly refractive threads of the eyes of insects upon the light, I made some experiments on the transmission of light through fine threads of glass.

I took a capillary tube of glass of an inch in thickness, about 1 of an inch in diameter, and an inch in length, placed it upright in a small trough of water under the microscope and examined it with an inch objective. I found that no light passed through the lumen of the tube, but that the section of the wall of the tube appeared brightly illuminated. I next placed a few fine glass threads, drawn from a glass rod, in the interior of the tube; these were as nearly as possible the same length as the tube and measured 1000 of an inch in diameter. The upper end of each of these rods appeared as a brightly illuminated disk in the dark field; when the focus of the microscope was altered the disk enlarged, showing that the rays left the rod in a divergent direction; in some cases when the ends of the rods lay beyond the focus of the microscope, the disks of light exhibited grey rings, the result of interference.

When the lower ends of these rods were lenticular, or fused into a drop, or drawn into a cone, the phenomena were the same, and in all cases the action of an oblique pencil, even when the obliquity was very slight, was feeble as compared with that of a pencil having the direction of the axis of the rod.

These results are such as would be predicted on the undulatory theory; all the light passing into the rod, except very oblique rays, would be totally reflected, without any change of phase in the undulations, at the surface of the glass, whilst all except the axial rays would be very much enfeebled by numerous reflections and interference from the different lengths of the paths of the rays. I think threads of a highly refractive character immersed in a medium of a less refractive index, when less than of an inch in diameter, would destroy the effect of rays of only very small obliquity by interference.

In order to determine the effect of the pigment, I covered the exterior of some glass rods of of an inch in diameter with black varnish, and I then found it impossible to transmit any rays of even the slightest obliquity through half an inch of such a rod.

From these facts I think it may be concluded that it is probable that the highly refractive structures may be regarded in the light of

VOL. XXVII.

T

luminous points, which serve as stimuli in exciting the recipient protoplasm in which their ends are imbedded.

The focus of the facet when this is lenticular, in all the insects which I have examined, is situated considerably deeper than the outer end of the rhabdion and below the surface of the rod cells in the microrhabdic eye, so that even for objects as close as of an inch to the cornea, we have to deal with convergent rays and not with a focal point. This indicates some mode of nerve stimulation other than the union of homocentric pencils, in a point beneath the compound cornea in relation with the recipient elements. Considering the small size of the parts, I think it quite possible that we must look to the phenomena of interference for the explanation; at least, they must play an important part in the phenomenon.

Whatever may be the manner in which vision is accomplished, the size of the corneal facets and the general curvature of the cornea renders the theory of J. Muller highly probable. It is true that Claperède has expressed the reverse opinion, but I shall endeavour to show that he has done so on insufficient data. According to his calculation, a bee should be unable to distinguish objects of less than eight inches in diameter at a distance of twenty feet from it. This calculation is based on the idea that the acuity of vision in this insect is the same in all parts of the field of vision, and that the general surface of the common cornea is approximately a segment of a sphere. This is not the case, for the angles subtended by the adjacent facets in the centre of the cornea, which is considerably flattened, is not more than half a degree at the most; so that on J. Muller's theory, supposing each facet to give rise to only a single luminous impression, the bee should be able to distinguish objects of about two inches in diameter at a distance of twenty feet, an acuity of vision quite equal to account for all the phenomena of vision in bees.

I have measured the curvature of the cornea of a number of insects, with a view to determining the angles made by the lines of vision drawn from the centre of adjacent facets. This is done in the following manner :- -A magnified image of the cornea is thrown on a sheet of white paper, by means of a microscope and camera lucida, and the curve of its profile drawn; in this way I have found the principal meridians. These curves approach more or less closely to an epicycloid.

It is easy with such curves and the size of the corneal facets to determine the angles made by adjacent facets. The angles vary inversely as the radius of curvature, and, therefore, the acuity of vision varies directly as the radius of curvature when the diameter of the facets remains the same, and inversely as the diameter of the facets when these vary in size. In many insects, as Tabanus, the peripheral facets of the cornea are twice or three times the diameter

of those in the centre, and the radius of curvature is very short at the extreme periphery.

In most insects the acuity of vision determined in this manner diminishes very rapidly at the periphery of the field. In the centre of the field it enables them to perceive, as distinct, objects which subtend one degree. In Eschna grandis the sharpness of vision is much greater, as the adjacent facets make an angle of only eight minutes with each other. This is the least angle I have measured in any insect, but I have no doubt, from the nature of the curve forming the meridians of the eye in the great dragon flies, that a small part of the centre of the field has a much greater acuity of vision than this; in the wasp the angle subtended by the smallest visual perceptions is twice as great as in Eschna; and in the bee it is half a degree.

The direction of the visual line, or the line perpendicular to the compound cornea in the centre of the field of most acute vision, varies in different insects. In the predaceous kinds it is directed forwards in the plane of the body, or forwards and outwards, making an angle of 30° between the visual lines of the two eyes. In the pollen feeders it is directed downwards as well as forwards and outwards.

The size of the corneal facets varies in different insects from

2000 to 750 of an inch in diameter. Their size, except in a few insects, is dependent on the size of the insect, the largest insects. having the largest and the smallest the smallest corneal facets. From this it follows that the vision of large insects is more perfect than that of small ones, except where the curvature of the cornea is very flat. This corresponds with the manner in which the insects fly. For instance, the small Diptera fly round in small circles, and seldom leave the place in which they first attain their adult condition, except when borne away by currents of air, whilst the larger species take long flights when disturbed or in search of food. The experiments of Muller and others have shown that the direction and length of the flight of insects depends largely on the visual powers of the insect. The forward flight of Tabanus and of many flies corresponds with the direction of their visual line, and the same may be said of the lateral movements of the large dragon flies.

The mimicry of insects, especially that between the Diptera and the Hymenoptera is sufficiently close to be a protection or advantage to the unarmed insect, and is such that it would render the one indistinguishable from the other, or the two insects would be scarcely to be distinguished under conditions of vision equal to those with which the insects appear to be endowed except at very close quarters.

In the extreme periphery of the cornea the adjacent facets make an angle of from 30' in wasps and some other Hymenoptera, to 12° in many insects. In the microrhabdic eye of Tipula the curvature of the common cornea approaches the segment of a hemisphere.

In most insects the field of vision has a small region common to the two eyes in the vicinity of the mouth; it is chiefly developed in the predatory species, and probably serves in determining the distance of their prey from their mandibles.

III. "Measurements of Electrical Constants. No. II. On the Specific Inductive Capacities of Certain Dielectrics." By J. E. H. GORDON, B.A. Camb. First Series. Communicated by Professor J. CLERK MAXWELL, F.R.S. Received March 9, 1878.

(Abstract.)

The author has, under Professor Clerk Maxwell's directions, carried out some measurements of specific inductive capacities by a new method. The essential features of it are:-

(1.) It is a zero method.

(2.) The electrified metal plates never touch the dielectrics.

(3.) No permanent strain is produced or charge communicated, as the electrification is reversed some 12,000 times per second.

The potentials of the electrified plates were about equal to that of 2,000 cells.

The following are the results obtained :-The solid dielectrics were plates 7 inches square, and from inch to 1 inch thick.

*These results are corrected for cavities in the plates. The mean of the uncor

rected determinations is 1:4864.