208 
of computer limitations. The strip triangulation was originally programmed 
(1953) for the IBM 650 which contained a 2000-word drum. Later the program 
was used with a simulator on the IBM 1620. Increasing use of analytical 
triangulation shown by the demand for copies of the program required fur 
ther revision and its large scale production. A FORTRAN version with a 
complete description is given in Schut (1973). 
Duane Brown, earlier an associate of H. Schmid, has made major 
contributions to the analytical treatment of aerotriangulation as also in 
various engineering problems. Brown's principal contributions are the 
following: (a) Treatment of all orientation parameters as either known or 
unknown; (b) Solution of normal equations achieved by partitioning their 
matrix to separate the orientation elements and ground points; (c) The 
method of introducing ground control points and the air-station parameters 
with appropriate weights, thus making it possible to include auxiliary data 
without, however, disturbing the basic mathematical model; and (d) De 
velopment of a new mono-comparator which works on the principle of self 
calibration (Brown 1976). 
Certain important contributions were made at Cornell University 
under the guidance of Arthur McNair by Anderson (1964) and Mikhail (1963) 
in developing the Triplet method (Anderson and McNair 1966). In this 
method the rigid elemental unit for the strip or sub-block formation is 
obtained by using three consecutive photographs (hence the name Triplet). 
Each triplet is overlapped with the adjacent one. The method was adapted 
for official use and yielding good results by the US Coast and Geodetic 
Survey (currently NOS, NOAA) (Keller and Tewinkel 1966). 
Through the group studies on Analytical Block Triangulation spon 
sored by ISP Commission III during 1964-68 it has been shown (Ackermann 
1968; Ghosh et. al 1968) that (1) Excessive sidelap (e.g., 60% as against 
20%) does not yield much improved accuracy in block adjustment; (2) Con 
trol in the periphery of a block (at least in the corners) would greatly 
improve the accuracy of block triangulation; (3) Additional auxiliary 
data as additional control in the central area of the block would also 
improve the accuracy; (4) The precision is practically independent of the 
block size but is directly related to the available control, its quality 
and distribution. 
By the end of the 1960s, we reach a stage w T hen simultaneous ana 
lytical block triangulation reached a level of maturity. Comparators 
(both mono- and stereo-) of various manufactures and designs came on the 
market, powerful computers were available and usable economically as well 
as complex and refined programs for computations and adjustments were 
developed. The simultaneous procedures known also as "Bundles" method were 
improved and adopted by many organizations (Matos 1971, Wong and 
Elphingstone 1972, Schenk 1972) and at numerous centers in Germany, 
Finland, Italy, Canada and the USA. In spite of these developments, how 
ever, the method of independent models remains very popular and is general 
ly found to be more cost-effective. In this, the solution of the normal 
equation system has been found to be critical as far as the preparation of 
the computer programs is concerned. A direct method by using submatrices 
as units and a Choleskv solution was adopted finally. This method has 
been called Hyper Cholesky (abbr. Hychol) and it has proved to be suitable