UM n 
EN - 
vw 
  
base its judgments. The errors of observation, of recording and the various distort- 
ions produce a complex pattern of errors which has to be examined at each stage 
of the calculations of a strip. It should be a very difficult task to substitute the 
intelligent inspection of the results of calculation as they build up by a code of 
reasonable length. 
To have all the calculations relevant to even a single strip consisting of a few 
photographs carried out fully automatically is hardly a practical proposition for 
yet another reason: it would require too large a capacity for storing data. The alter- 
native of using the electronic computor to calculate each stage of the triangulation 
separately requires, for an economic use of the machine, the piling up of a large 
amount of data. 
The strength of the above arguments may of course reduce with developments 
in electronics; the undisputable fact is that electronic machines are not available 
for many survey establishments, and that they are more suited to other types of 
calculation. 
2.2. The Use of Desk Computing Machines. 
On the other hand, it was felt that the usual type of desk calculating machines 
leaves much to be desired. The modern types of desk calculators are both efficient 
and fast. In fact, the working time of the machine is only a small fraction of the 
total time consumed in routine computing. Most of the time is consumed in (a) 
setting numbers on the machine; (2) checking the setting; (c) reading, booking and 
checking the booking of intermediate results; (d) setting the intermediate results 
in later stages of computation; and, (e) carrying out check calculations which 
often amount to duplication of the whole work. The need for these check calcu- 
lations arises primarily from the liability of the operator to make mistakes in the 
operations enumerated above. 
It appears therefore that a computing equipment which can make better use 
of the speed attainable by electric desk calculators, and at the same time remains 
inexpensive and easy to handle and maintain, might well offer a solution to the 
problem of computing in analytical photogrammetry. 
The preceding analysis of the time consumption in calculating with desk 
machines shows that effective reduction in time and effort would be possible if 
means were provided to satisfy the following requirements at low cost: 
(1). to store numbers in such a way that any of them can be set automatically on 
the machine registers when needed; 
(2). to transfer automatically the results of a calculation to the setting or multi- 
plier's register as may be required in chain calculations; 
(3). to store an intermediate result for future use; 
(4). to store an arbitrary sequence of operation so that it can be performed with 
a minimum of intelligent effort. 
It has been found possible to design a simple device which would make calcu- 
lating machines of the desk type satisfy the requirements enumerated above. The 
device is essentially a simple register on which the order of performance of a 
number of electrically controlled operations can be set and stored, temporarily or 
permanently, in such a way that the operations can be conducted automatically in 
the stored order. It may be better to confine description to the storage of the 
3