708 

very straight and steadily in order to cover the ground 
correctly, and, when one is flying straight and steadily, 
simple vertical indicators such as the spirit level 
indicate truly. 
Now it had been found, in connexion with bombing 
experiments during the War, that it is comparatively 
easy to fly very straight and steadily when no other 
condition is imposed, but that it is much more difficult 
to do so when trying to pass over a pre-arranged point. 
The reason for this is that the pilot normally estimates 
the horizontal and vertical by reference to the horizon, 
and, when forced to look at the ground beneath him, 
is quite unable to distinguish the true vertical from the 
apparent vertical as distorted by acceleration.? It is 
obviously necessary to look at a point in order to get 
over it, so that the reason for the distinction between 
merely flying straight and flying straight over an object 
is at once apparent. To overcome this difficulty it is 
necessary to devise methods of carrying out the survey 
with the minimum attention to the ground beneath. 
We decided to divide our experiments into two groups 
as follows : 
1. To study the accuracy with which it is possible 
to keep the camera at a constant height with its axis 
vertical, when the difficulty of finding a predetermined 
track is reduced to a minimum. 
2. To find out how to fly over a predetermined 
straight track without losing accuracy from that 
determined above. 
In connexion with the first group of experiments, we 
were fortunate in having near Cambridge a stretch of 
very flat country covered with numerous well-mapped 
and easily identified points and traversed by two large 
straight canals, more than twenty miles long, called the 
Bedford Levels. 
It is easy to fly down a long straight landmark of 
this nature with very little attention to the ground 
immediately beneath. We therefore flew along these 
canals at about 10,000 feet, keeping as straight and 
level as possible, and taking a series of photographs at 
regular intervals. We then developed these photos 
and, from them and a 6-inch Ordnance Map, calculated 
the position and orientation of the camera at the 
moment of exposure by the method of re-section. - 
The re-section was very laborious, but eventually, 
after about two years’ work, we obtained results for 
170 exposures, and these showed a probable error of 
tilt of about 1° from the vertical and a probable varia- 
tion of height from the mean of a flight of go feet. The 
distribution of errors in both tilt and height was quite 
normal. 
From this data it is clear that the tilt of the camera | 
axis from the vertical seldom exceeded 2° and that 
the height of the camera seldom varied more than 
roo feet from the mean of each flight. Simple calcula- 
tions, supported by previous experience in Palestine, 
lead to the conclusion that such errors should not 
introduce serious errors into maps made on the assump- 
tion that the axis is vertical and the height constant. 
This excludes, of course, errors that are cumulative over 
large distances. 
2 Tt is easily shown that a pilot, looking down at a point beneath him and 
trying to fly so as to pass over it, will tend to fly along one of a series of 
curves of which the equation is fp=const., where / is the perpendicular 
from the origin on the tangent from the point where the radius of curvature 
is p. These curves in general never pass over the required point (i.e. the 
origin). 
NO. 2795, VOL. IIT | 
NATURE 

[May 26, 1923 
COVERING THE GROUND WITHOUT Gaps. 
The second problem was to cover the required 
country to be surveyed without leaving gaps and with- 
out losing accuracy. It is on this problem that most 
attempts at commercial surveyin® have broken down, 
the primary cause of failure being the inability of the 
pilot to distinguish between the true and apparent verti- 
cal. So long as the pilot is allowed to look constantly 
down at the ground, in an endeavour to cover it accur- 
ately, tilts up to six or more degrees are liable to occur, 
and the tracks that are made under these conditions 
are often so curved as to cause large gaps between the 
strips of photos. 
By experiment we have found that the best way to 
solve the difficulty lies in allowing the pilot to locate his 
position by reference to the ground beneath, at the start 
of each flight only, and insisting that he must fly hence- 
forward without further reference to the ground. 
To do this in such a way that successive flights on 
the out and return journeys will cover the ground in 
parallel strips, it is necessary, first, to find and allow for 
the wind at the height in question. This we do by a 
method that was developed by the Air Ministry for 
purposes of aerial navigation, and we have brought the 
routine to such a pitch that within ten minutes after 
reaching the survey height, 10,000 feet, we can find the 
wind and make all necessary calculations for compass 
courses, etc. It requires considerable training and 
experience before this can be done. 
However good the methods employed, the strips on 
successive journeys will not be exactly parallel, and, 
since they are located only at one end, their length will 
obviously be limited if gaps are to be avoided. We 
find that this limit comes out at between ro and 15 
miles. The starting points for the strips are either 
marked on existing maps or on preliminary strips of 
photographs, taken along the edges of the mosaic at 
right angles to the mosaic strips ; these preliminary 
| strips are called “indication strips.” 
The pilot, therefore, begins by getting over a point, 
as accurately as he can, and then taking up a pre- 
calculated compass bearing as quickly as possible. A 
difficulty may here be experienced owing to the well- 
known fact that compasses on aeroplanes are affected 
by a change of course and do not settle down on a true 
| bearing until the aeroplane has been flying straight and 
steadily for some time. If the pilot has managed to 
get over the starting point while flying on the correct 
bearing, this difficulty will not arise, but when working 
from tractor aeroplanes, as we are forced to do, one 
cannot always manage this, because the lower plane 
obstructs the view of the point during the approach, 
unless the approach is made in a curve. 
To overcome this difficulty we use an apparatus 
designed by the Royal Aircraft Establishment. This 
consists of a free gyro that can maintain its orientation 
in space for some ten minutes, without reference to 
the movements of the aeroplane. We release this 
gyro while flying on the required course, just before 
reaching the starting point, and use it to return to 
the correct course immediately after passing the point. 
We consider that an apparatus of this nature will 
always be a great help in aerial surveying, especially 
when working from tractors, but we think that a 
: 
: 
: 
| 
: 
