(b) The position of a horizontal plane in the stereoscopic image which can be determined from a comparison of two vertical angles measured on the ground with the corresponding angles on the stereoscopic image measured in relation to an approximate position of the horizontal which can afterwards be corrected.
The data so obtained set the machine for (a) The length of the air base.
(b) The angle through which both photographs must be rotated about the air base, in order to re-establish the vertical plane con taining the air base.
(c) The inclination of the air base to the horizontal.
However the plates of a pair have been set, detail on the relief or stereoscopic image from a pair of set photographs is plotted by the superimposition of the image formed by a pair of artificial "floating marks" which combine stereoscopically to produce that image at a known position in space. The movement of the image relative to the landscape enables the latter to be measured and surveyed in detail. For convenience however in mechanical and optical construction it is usual for the "floating mark" to be fixed and the stereoscopic image of the landscape to be moved. Maps so produced have been mainly cadastral and engineering plans at a large scale. Meanwhile for topographical work at smaller scales methods have been evolved which depend upon the elimina tion, by good and consistent flying, of any excessive tilt. For all classes of air survey it is most important that navigation in the aeroplane should aim at maintaining an undeviating course both as to direction and height.
Methods for securing straight flight and constant height have not yet been standardised. As a rule each pilot has evolved his own methods for himself. Great skill is often displayed but varia tions of wind may well cause displacements of track of half a mile in a ten-mile strip. Deviations of 8o to too ft. in height are common. This problem, which can only be solved by a judicious mixture of instrumental control and training, remains one of the most important to surmount. If the tilt is small the perspective centre and the photo plumb point approach each other and the principal point. It is possible then to assume that directions from the principal point may hold good both as regards tilt distortion and height distortion. Methods based on this assumption are known as radial methods. The first assumption made is that the air base of any pair is defined in direction by the line joining the photographic images of the two principal points. It would not be possible however to accept as relatively correct on the ground the successive distances measured on the photographs between principal points. It is usual in these methods to allow for an overlap large enough to include the two air bases (forwards and backwards) from the centre of each individual photograph. The
positions of the fore and back stations can thus be found by a comparison of detail, but such tilt and height distortion as may occur on the lines between them are sufficient to introduce serious errors of length.
The scale of the plot is, therefore, obtained by accepting an arbitrary length on the first photograph and by continually fixing points in the overlaps and refixing the principal points from them thus:—The base 1-2 on the first photograph is set along any line which will be convenient for subsequent plotting. A line Ia is now drawn to a (on the photograph, image of A on ground) and is accepted as an arbitrary length on the "plot." The second photograph is now set on line 2-1 and moved along it till the direction from 2 to A cuts the position a on photograph I. Thus the distance is is made the base for all subsequent work. From i and 2 directions are drawn to b and c (images of B and C) which appear on both photographs. As these also appear on the third photograph which is set on the direction 2-3 (between prin cipal points 2 and 3) the position of 3 (principal point of the 3rd photograph) can be resected.
Naturally the construction lines must be drawn on tracing paper, linen or celluloid (the latter is most convenient), the photographs being placed separately and successively under it. Once the positions of the principal points have been plotted all other points of detail may be fixed by the intersection of the photographic directions. A particular development of the radial method, commonly known as the Arundel method, employs a small "topographical" stereoscope in these first stages of plotting. The particular feature is the addition over each photograph of glass plates on which are etched diamond shaped grids. These grids fuze stereoscopically into a plane which may be moved vertically in relation to the landscape, for contouring, and which are also used to show when the pair is properly in correspondence. By this method the orientation of the air base can be established and marked on each photograph. Contouring in the Arundel method is carried out on alternate photographs. Each pair is set and examined in the topographical stereoscope on the assumption that they are untilted and exposed from the same altitude. The result ing stereoscopic image is naturally subject to errors in apparent slopes, the effect of which must be held in check by a fairly close height control. The plane formed by the diamond grids may be made to touch the ground at any place, the extent of movement being recorded on scales. Other heights may then be interpolated and when a sufficient control has been secured contours can be drawn in by eye on the actual photograph. They are subsequently traced in correct position on the celluloid plot.