mark pointed on the photograph (diapositive). The measuring 
mark could be moved about very rapidly by means of the joy 
stick. , When within about a half-millimeter from a discreet 
image, such as a PUG mark, the machine could be commanded to 
center automatically on the image, which it could do within 
two seconds of time and punch the coordinates on a card, 
precise to a fraction of a micron. 
Another similar device made by Gilleland Instrument Co. 
of Oakland, California, has recently been delivered to the 
Autometric Operation of Alexandria, Virginia. It is my 
understanding that two or three of these instruments were 
scheduled for delivery before the date of this Congress. 
Presumably either of these devices, as well as a few 
other contemporary designs, has the capability of measuring 
aerotriangulation photographs at the rate of 10 to 20 per 
working day. This may be especially significant for satellite 
work. At the moment, my office requires about five man days to 
measure one satellite-geodesy plate on the monocomparator: 
the semi-automatic comparator may perform the work in two 
hours. It is also interesting that the semi-automatic 
device may be free from observational bias, which seriously 
affects human operators in measuring to microns. 
' i 1 
Such semi-automatic comparators are quite costly. The 
price in the U.S.A. is about 1.5 times that for a Wild STK 
stereocomparator, including import duty, or about h times 
as much as a monocomparator. Autometric Operation must be 
commended for establishing a "precision mensuration laboratory" 
where one can arrange to have his photographs measured. The 
principle is similar to that of a computer center: whereas 
one firm would probably not have sufficient work to utilize 
the full capability of a semi-automatic comparator, a central 
laboratory can serve numerous small users without having 
the large financial investment falling heavily on any of them. 
Programming and Computing 
The one who develops an analytic aerotriangulation system 
is severely punished by the cost and effort for computer 
programming. To develop a modest program may require from 
three to six man-months of effort. The adoption or adaptation 
of others' programs at small cost in the future seems to be 
inevitable. 
Thus the subject of the program language becomes 
significant. In the U.S.A. it seems correct that the Fortran 
notation is most popular. It is disconcerting, however, 
that even Fortran programs differ and may not be operable 
on other machines of the same manufacturer. Happily, the 
IBM firm has now included completely compatible Fortran