During an extended experience of cloud photography, it was found that it was quite possible to get pictures which showed the cloud detail even when the sun was in the field of view. Sometimes the solar image was reversed, but if the exposure was very short this was not the case. In such photographs the structure of the cloud was exceedingly clear and sharply defined quite close to the sun. Indeed, the intense illumination seemed to reveal minute details of internal arrangement which could not be detected in similar clouds some distance away.

The methods which had been employed for the measurement of cloud altitudes elsewhere have already been briefly referred to. Some of them required two observers, who were equally responsible, each of them having to direct his apparatus or camera to the same point of the cloud, and to record the exact direction in which the instrument was pointed. The instruments, if accurate, were costly, and there were many opportunities for error in reading the graduated circles which gave the directions. Moreover, in most of these methods the two observers were connected by telephone, and had to agree on the exact point towards which their instruments should be directed; either the exact point of the cloud, or the precise direction as shown by the mounting of the camera or other instrument. At Kew some of these sources of error were avoided by fixing the two cameras with the axes of the lenses and centres of the plates in a vertical position and exposing the two plates simultaneously. The Kew observations were not long continued, and for some years the only measurements in progress were those carried out abroad, particularly at the Blue Hill Observatory and at Upsala.

The experience gained in photographing clouds in order to record their forms suggested a way in which many of the sources of error in previous measurements of altitude could be avoided, especially by simplifying and reducing the operations at the moment of making the observation.

If two cameras are placed at the opposite ends of a measured base line, whose direction is known, and if they are both pointed towards the sun, on making the exposures by electrical means at the same moment, the position of the image of the sun upon the plate gives the direction in which the cameras are pointed. It will be in the same direction as seen from both ends of the line.

Now, if we note the time at which the exposure is made, this with the date gives all that is required for ascertaining the sun’s position in the sky, and is, therefore, the only exact observation which need be made at the time of taking the photographs. Mistakes are almost impossible, as each plate contains its own record of the sun’s position, and even if some of the plates should get mixed the images of the clouds will generally suffice to pair them properly. For general measurements there is one grave defect in the method, and that is that it can only be used when the sun and cloud can be got into the same field of view. But with the higher varieties of cloud this is generally possible, and it was just these higher sorts about which knowledge was least certain, and which it was proposed to study.

An initial difficulty was the finding of a level site, flat land being very uncommon in Devonshire, but fortunately a suitable place was found in some artificially levelled ground close to Exeter, belonging to the London and South Western Railway Company. It was a stretch of ground intended to be covered with sidings, but had not been finished, and had become overgrown with grass, stunted sallows, and other wild plants. Being railway ground, it was, comparatively, though by no means entirely, free from mischievous and inquisitive people. The next point was a suitable camera. It must have fairly long focus in order to give a large image, and therefore large displacement; it must be capable of being pointed in any direction and clamped there; and it must be capable of standing considerable extremes of temperature and variations of dampness, as it was intended that they should be kept on the spot in wooden structures, which served for stands as well as to contain the apparatus.

The pattern finally decided upon is represented in Plate 58, which shows one of the cameras pointed up to the sky and standing on one of the stands. These cameras were to take plates of whole plate size, two double dark slides of the ordinary pattern being attached to each.

There is no focusing screen. Focusing was done with great care once for all, and then a coat of hard varnish was put over all the adjusting screws. A small view-finder is attached to one side, and it was by this that the camera was pointed in the desired direction.

In order to lessen risk of mistake, it was so arranged that the two slides belonging to one camera would not fit the other. The lenses, of 18 inches focus, and giving sharp detail all over the plate, were carefully matched, and the focus adjusted until the images given by them when placed side by side appeared to coincide exactly. They were provided with iris diaphragms, which were shut down to an aperture of a quarter of an inch, and with shutters which could be released at the same moment by an electric current, acting through the electro-magnet shown under the lens on the front of the camera.

The shutters were of the kind known as the “Chronolux,” which will give any exposure from the sixty-fourth of a second up to three seconds. But it was found in practice that the highest speed was sufficient and gave satisfactory results. Of course, there was no idea of adjusting matters on each occasion so as to get the best possible negatives capable of yielding good prints. Measurement was the object, and if the negative showed the sun and sufficient cloud detail for the identification of cloud points, that was all that was wanted. The shutters gave a good deal of trouble at first. Their sliding parts were made of ebonite, and when the cameras were left in their stands with an August sun shining down upon them, everything inside got very hot and the ebonite warped; but the difficulty was got over by substituting aluminium.

The two camera stands were placed 200 yards apart, and were connected by a line of telegraph wire carried on short poles. At each end of the wire an insulated connecting piece was brought down to the camera stand, and to the batteries and other apparatus. The current which was sent through this wire by pressing a contact at one end of the line did not directly make the exposures; but two similar relays were brought into action, and each of these sent the current from a local battery of LeclanchÉ cells through the electro-magnet on the camera and made the exposure.

After development the two negatives showed the image of the sun, not far from the centre of the field of view, and the cloud whose altitude was required. Since this was taken from two different points of view, the negatives were not alike, but the distances between the centre of the sun’s disc and any special point of the cloud were different. For instance, if the cloud were east of the sun, with its edge just apparently touching the solar image as photographed from the eastern station, then the negative taken from the western end of the base would show an interval of clear sky between the two, which would be greater as the cloud was lower.

It often happened that after developing the plates the image of the sun was lost in a black blur, but it was easy to reduce this part of the image by local application of a reducing agent[3] by means of a paint-brush, until the disc became clear enough. Two lines were then drawn on the negative, one vertical and the other horizontal, intersecting each other at the centre of the sun’s image. These lines served as the starting-points for exactly measuring the distance from their point of intersection to any selected point of the cloud.

The distances could generally be determined to a fiftieth or a hundredth part of an inch, and their difference was, of course, dependent upon the direction of the sun relative to the base line and the altitude of the cloud, but for low level clouds the difference was sometimes so great that no pair of corresponding points could be detected, while it was often as much as an inch. With higher clouds the differences were smaller, but unless the sun was very low in the sky, either east or west, the displacements of the cloud image were great enough to give reliable measures. Specimen prints from pairs of negatives are shown in Plates 59 and 60.

The processes by which the measurements are worked out are laborious,[4] and consist of two parts, the first being the determination of the exact position of the sun from the date, hour, and latitude and longitude of the place, and the second, the determination of the position of the cloud. Two points which represent the same part of the cloud are selected, and their respective distances from the two lines drawn through the sun are measured as accurately as possible. Now, a certain distance on the negative corresponds with a definite angular displacement, and a scale can be constructed showing how much should be added to or subtracted from the sun’s position to get the exact position of the cloud. This being done, it is then a simple piece of trigonometry to deduce the actual height of the cloud above the place of observation. The work of computation, however, was greatly lightened by the fact that many of the pairs of negatives showed more than one layer of cloud; thus Plate 59, which is a fair specimen, shows three layers, and, consequently, one determination of the sun’s position sufficed for three distinct results.

For the highest clouds the displacements were, of course, small, and could only be made with certainty of a correct result within about three hours of noon. Earlier than 9 a.m., or later than 3 p.m., the sun was too nearly in a line with the two stations, or too low in the sky, to give a sufficient displacement of image. A base line of 400 yards instead of 200 would have been better for the high clouds. But, on the other hand, when low level clouds are viewed from two different spots their outlines may seem so changed that it may be impossible to identify a pair of corresponding points, and the same difficulty may also arise when high clouds are seen through a gap in a lower stratum. The longer the base line the more frequent and more obtrusive would this perspective difficulty become, so the distance of 200 yards between the stations was adopted as a convenient mean.

The method of making the observations was simple. Each observer was provided with some signal flags, by which the necessary communications were made in accordance with a simple code. Call the two observers A and B, and suppose A directed the operations. He watched the sky until a favourable opportunity seemed to be approaching. He then signalled to B, and both cameras were turned to the sun, the dark slides were inserted, the shutters set, and everything made ready. Signals were then interchanged, to signify that preparations were complete, and when A saw that the edge of the cloud had reached a suitable position to be in the same field of view with the sun, the contact key was pressed and the plates simultaneously exposed. At the moment when this was done the time was noted. Several observations were thus made in a short time.

Measurements were carried out as opportunity allowed over four consecutive seasons, from the beginning of April until the end of October. During the last of the four years, the site had become less convenient owing to an extension of the railway work, and early in November the series was brought to an abrupt conclusion by a heavy gale, which snapped off all the poles carrying the connecting wire. But by that time 423 measurements had been obtained, the great majority of which referred to clouds of the cirrus and alto groups.

The general results may be tabulated thus, giving heights in metres:—

These values are not very different, on the whole, from those which have been arrived at elsewhere, and in making a comparison it must be borne in mind that there is always a little want of precision in cloud nomenclature. As a whole, the Exeter maxima are greater than the foreign ones, and this is very markedly so in the case of cirrus, for which the American highest record is 14,930 metres, the Swedish record is 13,376, while the Exeter value is 27,413 metres, or about 17 miles. But this extreme measurement, and several others unusually large, were made in one morning, a day of very hot damp weather, when cloud formed at seven different levels: cumulus at a height of 1·9 miles, alto-cumulus at 3·9 miles, cirro-cumulus at 4·7 miles, cirro-stratus (No. 1) at 8 miles, cirro-stratus (No. 2) at 9·6 miles, cirrus at 11·5 miles, and cirrus excelsus at 17 miles. By about half-past one in the afternoon the sky was completely overcast with dull grey clouds, which cleared off at half-past four, and at half-past five in the evening the cirrus had fallen to 7·9 miles, and the cirro-cumulus to 4·3 miles. If this one day’s observations had been omitted, the Exeter maximum would only have been little more than 1000 metres above the record from across the Atlantic, but 1000 metres is a height worth noting.

While the Exeter maxima are all rather greater, we find the minima for cirrus, cirro-stratus, and cirro-cumulus are rather less than at the foreign stations; that is to say, that clouds are formed over Devonshire both at lower and at higher levels than seems to be the case in Massachusetts or Sweden. It seems probable that this is due to a greater humidity on our western coasts, such as we should suppose would be the case from their position and the prevailing winds and ocean currents. If so, we should expect the great convection clouds to be larger. Thus, at Exeter, out of only fifteen examples of cumulo-nimbus, the top varied from 2004 metres to 6409, with an average base level of 1045. At Upsala the maximum was 5970 and the minimum 1400, with an average base level of 1400. The mean thickness of the Swedish clouds was only 1400 metres, while that of the Devonshire specimens was more than 2000 metres.

Again and again, during the progress of these measurements, it was found that the greatest altitudes and the richest development of the higher varieties occurred towards the end of a spell of fine calm weather, when convection had had free play day after day. A slight fall of the barometer, only the hundredth part of an inch, would usually, under those circumstances, bring about abundant formation of high clouds, frequently of the undatus kind. All the cumulus clouds, by which we mean to include alto-cumulus and cirro-cumulus, are most frequent when the levels of condensation are rising, while the stratiform clouds are an indication of no vertical movement or of active descent. Pure cirrus is indicative rather of movement in a horizontal direction, and may occur when the condensation levels are stationary, or when they are rapidly changing either way.

In broken weather the natural movements of the atmosphere and of its vapour are masked and disturbed by the strong eddies brought by the cyclonic systems. It not unfrequently happens that the region of disturbance does not reach up to the level of the highest cirrus, or, what is more probable, the cyclonic system leans so far forward that we may have in its rear the upper clouds floating quietly far above the comparatively shallow region of disturbance, while in front the upper part of the storm system projects above undisturbed air.

The frequent appearance of cloud almost at the same time at more than one level is at first rather difficult to understand, but it will be noticed that when this occurs the barometer almost invariably falls. Now, if we suppose that the air is nearly saturated at more than one level, and that the whole is then bodily relieved of some of the superincumbent mass, so that the barometer falls, the mass of air will at once swell up, being cooled from top to bottom simultaneously, and wherever it is damp enough cloud will be formed.

The converse is equally true. If we have cloud at several levels, and the whole is compressed by the addition of more air above, which is the case when the barometer rises, that compression will be accompanied by the generation of heat and the consequent disintegration and disappearance of the clouds.