A Method for High-power Stereoscopic Microscopy

When an object is viewed with both eyes, each receives a slightly different picture on the retina. These two views are fused into one by the brain, and the resulting composite view gives a strong impression of solidity and depth, dependent on the minute discrepancies in the two views. This appearance of depth can be duplicated by any instrument, such as binoculars or a stereoscope, which presents suitable slightly different views to the two eyes.

In microscopy, it is often important to determine the structure of an object in depth, so that stereoscopic microscopes would be valuable. It must be appreciated that, whereas all stereoscopic microscopes are necessarily binocu-lar, not all binocular instruments are stereoscopic, as will be seen.

Binocular microscopes are of three main types :

1. Twin microscopes. These have two objectives as well as two eyepieces, giving excellent stereoscopy, but the two objectives limit them to low powers. Dissecting microscopes are of this type, with right-angled prisms in each tube to erect the image and provide adjustment for the inter-ocular spacing.

2. Beam-splitters. These have only one objective, the light from which is divided into two identical parts by the half-silvered surface of a cubic prism, or a similar arrangement. Both eyes receive the same view exactly, so the microscope is not stereoscopic. It is intended purely to relieve eyestrain during prolonged observation. The method is applicable to any power of objective or eyepiece.

3. Wenham prisms and variations. These also have only one objective, but prisms are arranged in the back focal plane of the objective so that the light from the left half of the back lens goes to the right eyepiece and conversely. Therefore true stereoscopy results, but at the expense of resolution, as the aperture of the objective is halved. It is relevant to point out that the stereo-scopic microscopes which work by employing a beam-splitter and placing opaque semicircular stops above the eyepieces in the Ramsden circles may be regarded as belonging to this class of binocular, and the same remarks about aperture apply. This is because optically there is no difference between stopping the Ramsden circle and stopping the back focal plane of the objective, since these are conjugate planes. Owing to the drop in resolution due to the decrease in the aperture of the objective, these systems are not suitable for the highest powers of the microscope.

It can be seen, therefore, that there is no well-known microscope giving true stereoscopic vision at high resolution. The purpose of this paper is to reintroduce a way of filling this gap.

This method was suggested in a short paper by Jackson (1944−9), which unfortunately contained little detail and no explanations. The following is an expansion of his account with added remarks of my own.

The apparatus is a simple modification of a beam-splitting microscope— that is, one in which both eyes see identical views, obtained with the full objective aperture. Polaroid is inserted in such a way that the full objective aperture is used, only the condenser aperture being halved. As is known (Baker, 1952), this has a much less deleterious effect on resolution than halving the objective aperture, as in the Wenham systems.

A disk of Polaroid is cut, of the correct size to fit into the substage filterholder. A line is scratched lightly diametrically across the disk and parallel to its plane of vibration. This may be determined by crossing it with a Nicol or a marked piece of Polaroid. The disk is then cut diametrically across, at an angle of 45° to the plane of vibration. One half is turned upside down and the Polaroid is mounted dry between disks of glass and bound around the edge with adhesive tape. The two halves will then be polarized at right angles to each other, and at 45° to the junction line. Alternatively, the disk may be constructed to give vertical polarization in one half, and horizontal in the other. In this case, one of the two possible orientations of the disk in the microscope will probably give better results than the other.

The binocular beam-splitting microscope is set up in the normal way, with a powerful light source (an opal bulb is inadequate). I find that a Pointolite gives excellent results. The Polaroid disk is then placed in the filter-holder immediately below the condenser, in such a position that it is divided into left and right halves of opposite polarization. This is checked by observing the back lens of the objective: if the two halves are not equal, the disk is not centred, and this must be rectified.

Two Polaroid analysers are placed in the two eyepieces, preferably just above the field lens, so that specks on them will be out of focus. The Ramsden circles above each eyepiece are now examined, either by drawing the head well back from the microscope, or with a hand-lens. They are conjugate with the back focal plane of the objective and the lower focal plane of the condenser, and an image of the polarizer is seen: each Ramsden circle is there are divided into left and right halves. The eyepieces, with analysers, are rotated until the right half of the left Ramsden circle and the left half of the right circle—that is, the inside halves—are extinguished (fig. 1, A). True stereoscopy then results, which may be checked by comparing the views of the two eyes. If the analysers are rotated so as to extinguish the outside halves of the Ramsden circles instead (B), pseudoscopy or reversed stereoscopy is obtained; while extinguishing both left (c) or both right halves, or equalizing the intensity of the two halves of each disk (D), gives non-stereoscopic o inocular vision.

The method is applicable to any power of the microscope, and objectives can be changed without any altering or resetting. In fact, it has been found that the 2-mm lens gives the most startling results, no doubt owing to the great angular difference between the two direct light beams. It is thought that the failure of the method to achieve popularity in the past was due to its being tried only with low-power objectives, with which its performance is much inferior to a Wenham prism. The advantage of the method lies in the strong stereoscopic effect produced with a 2-mm objective.

Light normally strikes a microscopical preparation as a cone, symmetrical about the optical axis. If the light is made oblique, e.g. by moving a substage stop, the object appears to change its position and angle to the optical axis by an amount proportional to the obliquity of the light. This must be due to a progressive change in the phase of the direct light from one side of the object to the other, therefore modifying the phases of the various components of the diffracted light in the back focal plane of the objective, and so altering the image. The divided polarizer provides two oblique half-cones of light which can be isolated by the analysers, and also two sets of diffracted light, each of full aperture since the objective is unobstructed, which can be similarly isolated.

The Ramsden circles confirm this—in each can be seen the bright half, which is the direct light of half aperture, together with half of the diffracted light. The dark half of the Ramsden circle is not completely black, partly because of imperfections in the Polaroid, but also because it contains the other half of the light diffracted by the object, and having the same polarization as the direct light. Each eye therefore receives direct light of half aperture, together with diffracted light of full aperture. If the object is strongly diffracting, it will be found impossible to darken half of the Ramsden circle because of the intensity of the diffracted light, and it will be necessary to move the object out of the field of view before the eyepiece analysers can be set.

It should be mentioned that this method cannot be used where the object is strongly biréfringent, though slight birefringence, as in muscle, seems to make no appreciable difference.

Depolarization by reflexion in the binocular prisms is not serious in any system that I have tried out, as a very complete extinction is not necessary, provided that the unwanted beam is small in intensity compared with the wanted one. However, some prism systems cause rotation of the plane of vibration, so the setting of the analysers is not necessarily the one expected. This was not remarked upon by Jackson, and is the reason for setting the analysers by observing the Ramsden circles.

Phase contrast can be used with this method of stereoscopy, although a very intense light source is required, and the stereoscopic effect is not so pronounced, because the depth in which the phase effect is produced is smaller than the depth in which out of focus detail can be appreciated stereoscopically. The stereoscopic effect of the focused detail is just as great with phase contrast as with direct microscopy.

I have also successfully applied patch stop dark ground illumination.

I am indebted to Dr. J. R. Baker and Dr. S. M. McGee-Russell for a most interesting discussion of stereoscopy, to Mr. T. A. Minns of Messrs. W. Watson & Sons for some references and comments, and to the Medical Research Council for the grant under which I am working.