109 
should not be generally applied. Taking into account the general spirit in which international tests 
on air-surveys and aerotriangulation (so called: essais contrôlés) were carried out by I. S. P. 
between 1952 and 1956, where each responsible expert was left free to apply his own methods and 
equipment, M. Cruset suggests international cooperation to be established as follows: international, 
national and private laboratories to exchange lenses and cameras so as to have them tested by 
different means and methods. Results obtained to be ultimately published by Commission I. Such 
results obtained in laboratory-tests should afterwards be checked by trials in airplanes. 
Je donne la parole au Dr. L. E. Howlett du National Research Council of 
Canada. 
Dr. L. Howlett: The trouble with this subject, on which so much nonsense is L. Howlett 
spoken, is: if you have a perfect lens, there is no problem at all. If you do not 
have a perfect lens, which you never have, then all the language which you 
apply to a perfect lens cannot be just transferred to an imperfect lens and make 
sense. Such terms as »principal point», »focal length» mean nothing to an ordi 
nary imperfect lens until one has done some definition. 
If we think in terms of calibration, the first thing that we want to define is 
the geometry of the system. Photogrammetrists are particularly interested in 
the principal point. Principal point defined by physicists for a perfect lens has 
meaning. But when we come to a lens in which there are both residual errors 
of design and manufacturing errors, you have none of these things. It is essen 
tial then to select certain quantities, focal lengths and principal point in such 
a way that what you have is as close as possible to what you would have had 
if the lens had been perfect. To do this, it seems essential to have a co-ordinate 
system. It seems to me, a rather happy thing, that when we discussed these 
matters at Washington, at the last Congress, mention was made of a prin 
cipal point of autocollimation. This at least gives a fixed point to which one 
can refer to distortion characteristics of a given lens. I think we were un 
fortunate in introducing the expression »principal point», since the principal 
point of autocollimation is not a principal point. It is merely a point that can 
be unambiguously defined and can form an origin for a system to which the 
distortion of the lens can be referred. In general, a lens has two distortions; 
one is commonly called radial distortion. This comes in the main from the 
residual aberrations left by the designer. It is always completely symmetrical. 
In addition to this distortion, another one is introduced by the manufacturer. 
I think it was an unhappy choice of word, that this distortion was called tan 
gential, since, in manufacturing, distortion does not have to take place in the 
tangential direction; it can, in fact, occur in any direction and cannot be pre 
dicted. Hence I think it would have been much happier if we had divided dis 
tortion into two neat parcels: one coming from the theoretical side, the residual 
left by the designer, and the other introduced by the inability of any manu 
facturer to manufacture a perfectly centred system. 
Now, when we get into the problem of the second error, we get into further 
difficulties in defining what we mean in calibrating. This brings up the problem 
immediately of whether you should calibrate visually or photographically. One 
presumes that a photogrammetrist wishes to have something which is meaning 
ful to him. In another work we recently have demonstrated that there is a 
difference, a measurable difference between the distortion of the Aviogon as 
measured visually and as measured photographically. This is a significant differ 
ence and I think it is the first occasion at which it has been measurable. It has 
been measurable because of the excellent definition which is obtainable by the 
Aviogon and, I think, obviously, requires that, to be meaningful, all future 
calibrations of aerial cameras must be done by a technique which is equivalent 
to the photographic method. 
Le Dr. Ilotvlett constate que la terminologie utilisée pour les objectifs réels et imparfaits 
s’applique mal à ces optiques puisqu’elle a été forgée pour l’objectif parfait, instrument qui ne pose 
aucun problème de contrôle. Il se félicite que l’on ait pu recourir à la notion de point principal 
d’autocollimation en ce sens qu’il fixé une origine aux mesures de distorsion, mais regrette qu’un