
May 26, 1923] 
NATURE 
797 

. Recent Experiments in Aerial Surveying by Vertical Photographs.’ 
ce 
re is proposed to describe in these columns the results 
_ of experiments on aerial surveying that have been 
in progress at the University of Cambridge since 1920. 
The experiments were made possible by the co-operation 
of the Royal Air Force and the Department of Scientific 
and Industrial Research with the chair of aeronautics 
at Cambridge. They were suggested in the first place 
by Mr. Hamshaw Thomas of Cambridge as the result 
of his experiences of air-mapping in Palestine during 
the War. The authors wish to acknowledge their 
debt to Mr. Thomas, not only for the original 
suggestion, but also. for valuable advice during the 
progress of the work. 

To make an accurate survey by air, it is necessary 
to have information concerning the position and 
orientation of the camera at the moment of exposure. 
If the ground concerned is flat and the iilt of the camera 
is known, the photograph can easily be re-projected to 
give a true plan. When the ground is hilly, two photos 
of the same area, taken from known points, and with 
known tilts, will provide information from which a 
complete model, or map with contours, can be con- 
structed. 
If three points that are accurately known in position 
occur in a photo, it is possible to find the position and 
orientation of the camera from internal evidence in the 
plate itself. This process is called ‘“ re-section.” It 
is thus theoretically possible to map an indefinite area 
of country from a single base of three known points, 
for these points could be made to occur in the first two 
photos, which could then be used to determine the 
positions of other points, from which further photos 
could be re-sected, and soon. In practice this process 
would lead to accumulations of error which, with the 
methods yet available, would soon become prohibitively 
large. It is for this reason necessary, if the re-section 
method is to be used at present, to provide a net of 
ground - surveyed control points such thac three will 
occur in most, if not all, the photographs. 
When the aerial map is merely required to record 
changes, or to fill in detail in a country that has already 
been closely surveyed, as, for example, in the war 
mapping of the Western Front, or in the re-mapping of 
towns in peace, many accurately surveyed points will 
already exist and the re-section method will be. the 
obvious one to use. When, however, the problem is 
to map large areas of unsurveyed country, as, for ex- 
ample, the interiors of Australia or Africa, the cost of 
providing so many ground-surveyed points will gener- 
ally be prohibitive. This will be especially the case 
when the country is flat and heavily wooded. 
It is, however, precisely in connexion with large 
areas of this nature that the outlook for aerial surveying, 
on a large scale, is most hopeful. In such cases it will 
not, in general, be practicable to spend much money 
per square mile of survey, so that it becomes necessary 
to employ methods that neither require a close pre- 
liminary ground survey nor involve too much office 
mead of two lectures delivered at the Royal Institution on February 
15 22. 
NO. 2795, VOL. 111] 
Eee 
’ : By Prof, B. Metvitt Jones and Major J. C. Grirrirus. 
labour per photo. Both these conditions rule out the 
re-section method for work of this class. 
Now exact determination of the position and orienta- 
tion of the camera in space, by methods that are inde- 
pendent of the photo itself, is a matter of great diffi- 
culty, but it happens that an exact plan of level country 
can be constructed from overlapping photos, withou: 
knowing the exact position of the camera, provided that 
all the photos are taken from the same height and with 
the camera axis vertical. The reason for this is that 
all such photos will show a true plan of the ground 
to the same scale, and therefore they can be shuffled 
together, until the detail joins up everywhere and a 
true plan is formed. 
If, therefore, a camera can be kept at a constant 
height, with its axis vertical, and moved about over 
the ground, so that the whole country is covered by 
overlapping photos, it will be possible to construct a 
continuous plan of the ground from contact prints 
siraight from the original negatives ; and it will not be 
necessary to provide for known points to appear on 
each photo, or to determine the position of the camera 
at each exposure. Moreover, the heavy office work 
involved in re-secting and re-projecting each photo will 
be avoided entirely. 
Such a process is strictly accurate only when applied 
to absolutely flat country, but when working’ from 
10,000 feet; as we do, undulating couniry up to about 
soo feet local differences of level can be classed as 
sufficiently flat from this point of view. It must also 
be remembered that flat country is often the most 
difficult to survey from the ground, and is therefore 
the country in which an alternative method is most 
required. 
We are thus led to the conclusion that the econo- 
mical mapping of moderately flat country, by means of 
vertical photographs, depends upon the accuracy with 
which the camera can be maintained at a constant 
height, with its axis vertical, and upon the ability of 
the pilot to fy so as to cover all the ground with 
photographs that will neither overlap too much nor 
leave gaps. The experiments at Cambridge have 
been concerned mainly with the accuracy obtainable 
in these operations, given suitable apparatus and 
sufficient training in air routine. 
FINDING THE VERTICAL. 
The problem of keeping a constant height is quite 
straightforward and easy to understand; the only 
difficulty is to do it. The problem of keeping the 
camera axis vertical is complicated by the fact that all 
forms of apparatus that are designed to indicate the 
vertical are disturbed by horizontal accelerations of 
the aeroplane, which often persist in one direction for 
so long as twenty seconds at a time. It is possible to 
devise gyroscopical instruments that will seek the 
vertical so slowly as to average out these disturbances, 
but much experience has been gained with instruments 
of this type during the War, and this experience was 
not encouraging, mainly owing to the liability to 
failure of the delicate apparatus required. 
In aerial surveying it is, however, necessary to fly 
