since the photographer had not anticipated the photogrammetric application of his photo- 
graph, he could not claim that the camera was exactly level at the time of exposure with- 
out the use of level vials. It was then necessary to explain that the camera did not have 
to be level at the time of exposure so long as the tilt of the camera was determined ana- 
lytically and this tilt was included in the subsequent computations. 
The third, and final, challenge to the metrical analysis of data from a non-metrical 
camera would have been correct had it not been for the loose accuracy requirements of 
this case. The exposure camera was a Speed Graphic with a coated Ektar lens. It was 
charged that the camera was not of rigid enough construction for metrical purposes be- 
cause of a movable lens mount. The camera was calibrated at the same setting as those 
for the accident exposure. Six calibration exposures were made: two with the top of the 
lens mount forced as far away from the image plane as possible, two with the lens in its 
normal position, and two with the top of the lens mount forced toward the image plane. 
The differences in the calibrated focal lengths from these separate lens positions were 
insignificant, and the average focal length was found to have a probable error of only 
0.26 mm which was within the accuracy limits of this case. 
A further complaint against the use of a non-metrical camera was that the film was not 
stable. Obviously film used for pictorial purposes may not have stable metrical proper- 
ties. Therefore, film distortion tests were made on several samples of the same type of 
film as that used for the exposure in question. The maximum film distortion in the corners 
was .08 mm. This would have been intolerable were it not for the low order of accuracy 
required. 
The plaintiff’s counsel continued to level attacks on various possible errors. He 
claimed that the foreground of the photograph was out of focus and looked like popcorn. 
This so-called ''out of focus'' was of course due to the depth of object space. The 
circles of confusion for the images of both ends of the skid mark were computed and 
found to be 0.003 inch for the near end of the skid mark and 0.001 inch for the far end of 
the skid mark, both of which are insignificant in magnitude in this case. It was pointed 
out that circles of confusion of the image of the road gravel immediately in front of the 
camera may have been large, yielding a ''popcorn'' effect, but were irrelevant since the 
area where they existed was not employed in the computations. 
A rather unique method was necessary to determine the skid mark length. An arbitrary 
datum in which the skid mark lies was determined analytically. The control data con- 
sisted of the elevations of car tail lights above the road and the distances to the tail 
lights from the camera lens. The elevations of the tail lights were measured on identical 
car models to the ones in the photograph. The distances to the tail lights from the camera 
lens were determined by plane resection equations. For these equations the angles sub- 
tended by the images of the tail lights and a midpoint on the cars between the lights were 
combined with the distances between these points, in object space, which were measured 
at the same time that the elevations of the tail lights were measured. These data were 
used as controls in determining the angular orientation of the camera with respect to the 
skid mark datum. The normal distances of the tail light images from the horizon (which 
is a plane through the lens parallel to the plane of the skid mark) were determined with 
functions of tilt and swing. An average height of the camera with respect to the arbitrary 
datum was then determined from six unique solutions. 
With the normal distance from the camera to the arbitrary datum, the functions of tilt 
and swing, and the camera coordinates of the skid mark terminals, the ground coordinates 
of the skid mark in the arbitrary datum were computed. The computed value of the skid 
mark length was 65.2 feet + a probable error of 3.1 feet. The photographic evidence would 
rest in favor of the driver since the skid mark length plus the probable error was less than 
the maximum allowable distance. 
Thus, due to a jury unfamiliar with the science, photogrammetry was burdened first 
with the necessity of justifying its position as an accredited metrical science, then 
photogrammetry was burdened further with the necessity of establishing its accuracy in a 
rather unorthodox application such as was necessary in this case. 
The plaintiff incorrectly labeled all unorthographic properties of the photographic 
image as ‘‘errors’’. Distinctions then had to be made between properties following 
precise geometric laws such as those of orientation and errors such as those of lens and 
film distortion. It had to be shown that photogrammetric properties of orientation could 
be included in further computations, and that the magnitude of actual errors could be de- 
termined and their existence admitted. Fortunately for the defendant in this case accuracy 
requirements rendered all errors insignificant. 
The photographic data here presented cost about four times what it would have cost if: 
1. The use of photogrammetry had been anticipated, 
2. A metrical camera had been employed, and 
3. There existed adequate photogrammetric instrumentation for photographic coverage 
of accidents. 
SOM 
Aer 
They ar 
and trai 
tive site 
given ar 
tos has | 
Tir 
because 
Th 
mixed 
northea: 
east. 
"Th. 
2 millio 
Th 
for a lo; 
some pa 
stocks. 
Th 
complet 
photogr 
MacDor 
the can 
interpre 
What H 
Ma 
decision 
decision 
this met 
the two 
then m 
some ca 
Dr. Spu 
readily | 
photogr 
use of tl 
and som 
solution 
Washing 
taneous] 
41