1940 - 1990

Joseph F. Dracup
Coast and Geodetic Survey (Retired)
12934 Desert Glen Drive
Sun City West, AZ 85375-4825

Page: 1 2

ABSTRACT

A whimsical, yet serious look at the events, technology advances and people of the past 50 years in geodetic surveying as seen through the eyes of one who traveled the road, all the way. The period is divided into three segments: Dawn of a New Era 1940 - 1950; Break with the Past 1950 - 1970, and New Age of Geodesy Begins 1970 - 1990. Each contains synopses of major happenings, anecdotes of the time and names of some of the people directly involved in the events. (C&GS) and (NOAA) following names indicate the persons were members of the commission officer corps.

DAWN OF A NEW ERA 1940 - 1950

The most progressive and productive era in American geodesy began innocuously enough, continuing much as it had in the 1930's and earlier decades. Coast and Geodetic Survey (C&GS) triangulation and leveling parties roamed the land establishing geodetic control, while a small office force processed the data for publication. This is how it had been for more than 100 years. There would be many changes in the next five decades.

Full Time Field Parties

Not surprisingly, field parties were to see the first of the changes. In the past, most personnel were hired for the length of the project and once the job was finished they were paid off and advised where and when another project was to begin. It was up to each to travel to the new location at their own expense. This practice changed in the 1930's and field parties became more permanent units, working the cooler climes in summer and the southern states in winter.

These arrangements made it conducive for party members to bring their families with them, living in privately owned house trailers in a self contained environment, usually in town parks, fairgrounds and the like. The all male tent encampments of the past, with their mess facilities, which contributed so much to the legends of field life disappeared, except for some mountain work and Alaska assignments. Field parties operated in this fashion for about 30 years before economic and other conditions made it impractical to continue that way of life.

First Computers Draw C&GS Interest

Events occurring at Iowa State University, Ames, IA, in the late 1930's and the University of Pennsylvania in the early 1940s would effect nearly everyone in the world in the ensuing decades. These events led to the construction of the first automatic electronic digital computer (EDC), the Atanasoff-Berry Computer (ABC) by John V. Atanasoff and Clifford Berry in 1942.

In 1946, J. Presper Eckert and John W. Mauchly under U.S. Army sponsorship, developed the mammoth Electronic Numerical Integrator and Computer (ENIAC) at the University of Pennsylvania. From this and other beginning efforts, John von Neumann developed the storedprogram EDC in the 1950's, the forerunner of present day equipment. The C&GS with its ever increasing volume of computations had a standing interest in anything new that might ease the burden and the developments didn't go unnoticed. Accordingly, early in 1942, Lansing G. Simmons, Chief, Philadelphia Computing Office made a visit to the University of Pennsylvania and found that the machine was still in a primitive stage and could not, for example, multiply positive and negative numbers and sum the results, an integral part of the adjustment process and geodetic computations in general. Nonetheless, he and others in the Geodesy Division were convinced that such machines would be the wave of the future, yet recognized it would be a decade or more before they would become available. In the interim period, high speed automatic computers were employed to ease the work load.

ADPs Used For Least-Squares Adjustment

In 1946, Charles A. Whitten made the first least-squares adjustment of triangulation using Bureau of the Census automatic, high-speed data processors. These International Business Machines (IBM) processors utilized punch cards in the calculations. The earlier computer developments had no direct bearing on this geodetic computation breakthrough, nevertheless the fact that the technology was on the horizon, was a contributing factor.

In 1948, the C&GS installed similar automatic computers (ADPs) and began the process of adapting them to geodetic computations. The ADPs were first employed to form and solve normal equations by the Doolittle method, up to and including the computation of the residuals stage. It wasn't that they could do the job faster, but more importantly that they could be run 24 hours a day, at a modest additional cost.

ADPs could solve about 15 equations in each 8 hour shift, and that was the solution rate most mathematicians (later changed to geodesist), in the Triangulation Branch, achieved in the same period. However, a few in the Branch developed superior speed and accuracy in operating rotary calculators, using a feel and sound system, rarely, if ever, viewing the keyboard or dials before completing a series of multiplications. This small group averaged 20 or more equations solved per day.

The end was drawing near for mechanical calculators, albeit it would be 20 years or more, before they were gone from most desks. More important to many involved in solving large sets of normal equations on such machines, was the fact that the drudgery of that job would soon be just a memory.

World War II Intervenes

During World War II, regular geodetic activities were suspended for the most part and much of the effort was directed to carrying out needed surveys at defense facilities in the U.S. and Caribbean area. Field parties operated at reduced strength or were disbanded and the office staff also was reduced as members of both groups entered the military. The major field work, accomplished in this period, was measuring an arc of first-order triangulation from Skagway in southeast Alaska, over White Pass to Whitehorse in the Yukon Territory of Canada, then following the Alcan Highway road builders north westward to Fairbanks, a distance of about 575 miles.

Alaska Surveys Accelerated

At the end of the war, with the connection to the lower 48 triangulation observed, plans were made to complete the primary horizontal control in Alaska and to tie the several local astronomic datums to the North American datum of 1927 (NAD27). This was a wise decision, considering that the Cold War got colder and oil was found on the North Slope shortly thereafter.

Several parties were in the field each working season and most of the work was done by 1956. Despite the hardships and danger, Alaska was considered a plum assignment and the per diem was higher too. The versatility of Bilby towers and the ingenuity of the men who built them was demonstrated time and again in Alaska.

Assistance to National Topographic Mapping Program

The last of the great arcs had been observed in the early 1930's, from Providence, RI to Key West, FL, following the Atlantic coast, a distance of about 1,600 miles. During the big push to complete the first-order network in that decade, which was funded by civil works programs, most were arc surveys. However, late in the decade the program to begin filling in the areas between the arcs was instituted. The program continued in the pre-and post-war years and into the following decades.

Much of the work was done in support of the U.S. Geological Survey's (USGS) nationwide topographic mapping program, involving control for a few quads here and a few quads there. This piecemeal approach was dictated by funding and other requirements. There was one exception, the USGS reached an agreement with the Commonwealth of Kentucky to accelerate the program there. To support that goal, the triangulation was completed, but for two small areas, by about 1950.

Although the area work was classified as second-order (the then 1:10,000 standard), the observations were made to first-order specifications, with minor exceptions. New standards and specifications were drawn up in 1957, which took these exceptions into account, so that all area nets could be classified to the same order of accuracy.

Computing Offices and SPCS

Computing offices were setup in New York City and Philadelphia to aid in adjusting the numerous new surveys made in the 1930's. Among the assignments, was the conversion of the published stations' latitudes and longitudes to plane coordinates on the recently created State Plane Coordinate System(SPCS). Early on, it was necessary to compute the required tables, as well.

These field offices were sponsored by the C&GS and funded by various civil works agencies. The NY Computing Office was in existence between 1932 - 1964 and the Philadelphia Office from 1940 - 1943. Both offices were placed under the Civil Service system about 1942. And, both contributed significantly to reducing the large backlog of work.

Calculating SPC's was a huge task, even after tables had been prepared. Originally both systems, Lambert and transverse Mercator, involved a combination of logarithms and natural functions and each computation was made in duplicate as a check. Tables for inverse computations were not available until the late 1940's. An addition check was made by computing selected grid distances between all points, converting these distances to geodetic values via scale factors and comparing them with the published results.

Even at the beginning, Lambert plane coordinates could be computed using natural functions, provided a 10-bank calculator was available, and few were in the 1930's and early 1940's. In the late 1930's Lansing G. Simmons, then in charge of the Georgia Geodetic Survey, developed tables based on empirical formulas to compute transverse Mercator coordinates using natural functions. The computation time was cut in half.

Use of Logarithms Near End

Many geodetic computations were made using logarithms, including the preparation of side and length equations, computation of triangle sides and geographic positions. The latter was a formidable chore, requiring 135 entries for longer lines. It was once estimated that the computation used up one mile of lead pencil. Others thought 0.5 mile was closer to the truth.

In 1941 Lansing G. Simmons, then in charge of the Philadelphia Computing Office, conceived position computation formulas, based in theory on the transverse Mercator projection, that was amenable to natural functions. Tables were prepared covering only the conterminous States (latitudes 24-50) and were in feet, with the idea that some surveyors and engineers would use geographic coordinates. It didn't happen.

And in 1944 tables were prepared in meters, extending from the equator to 75. In conjunction with the conversion, natural function tables for sines and cosines, at 1" intervals, were computed in the same office. The resulting Special Publication (No 231) soon became a best seller.

UTM Grid Developed

The Universal Transverse Mercator (UTM) grid, a world wide plane coordinate system was developed in the 1940's by the Corps of Engineers, U.S. Army, following the recommendations of Oscar S. Adams of the C&GS Geodesy Division. The grid consists of bands, 6 of longitude wide, and a maximum scale reduction of 1:2,500. Original tables (for the Clarke spheroid of 1866) were computed by a Civil Works project, in NYC, sponsored by the U.S. Lake Survey (USLS) during the early 40's. The USLS unit evolved into the Geodetic Division of the Army Map Service (AMS) about 1943. Later, tables were computed for other ellipsoids then in use. Floyd W. Hough, David Mills, Homer Fuller and Frank L. Culley were directly associated with the grid's development.

World War II Technology Advances Surveying

Technology developed during World War II began almost immediately to find places in surveying. Shoran, a special type of radar, was used to control hydrographic surveys in 1945. In 1946, Carl I. Aslakson (C&GS) examined its possible use in locating islands, atolls and the like, that were beyond visual means to geodetic accuracy. In 1951 the method was employed, with the longer range C&GS developed Electronic Position Indicator (EPI), to position 4 islands off Alaska in the Bering Sea.

In the early 1950's, the U.S. Air Force measured a Shoran trilateration net between Florida and Puerto Rico, continuing on to Trinidad and South America. A similar net connecting North America and Europe via Greenland, Iceland, Scotland and Norway was observed between 1953-1956 by the same organization.

In 1948 a novel survey procedure, flare triangulation, was used to connect the Florida mainland with the Bahamas. Flare triangulation involves simultaneous observations from at least 3 ground stations, on each land form, to a flare dropped at a prescribed location by a high flying aircraft. It was first used in 1945 to connect the triangulation of Denmark and Norway across the 90 mile span of the Skagerrak. While the experiment was viewed as a success there were too many problems, including the need for near perfect weather conditions, for the method to be generally accepted.

Birth of EDMI

Erik Bergstrand's experiments, in Sweden, to employ light to measure distances came to fruition about 1948. This event was to have the same dramatic impact on geodetic surveying as did Jesse Ramsden's direction theodolite, reading to 1", in 1787. This instrument, named the Geodimeter, would reduce the time required to measure base lines from weeks to hours, without any reduction in the accuracy of the line. Furthermore, it permitted the measurement of regular length triangulation lines, doing away with the costly and accuracy lessening expansion nets.

A very eventful decade thus ended, holding great promises for the future.

BREAK WITH THE PAST 1950 - 1970

High Water Mark

The golden age of geodetic field operations in the U.S. began shortly after the war ended in 1945 and reached its zenith in the late 1960's. It began to fade in the early 1970's as governmental reorganizations and other factors, including economic conditions, directed funding priorities elsewhere. Despite these annoyances, even a name change, after more than 160 years the quality of the work was not effected. If anything it became better, albeit the productive rate was reduced.

For about 20 years, numerous parties were in the field carrying out mostly triangulation and leveling with smaller units doing triangulation reconnaissance, astronomic and gravity work. Sub units were assigned to various space and DoD facilities including Cape Canaveral, Vandenberg AFB, Pt. Mugu and White Sands.

The optimum steel tower party had 3 observing units and about 40 personnel, similar mountain parties about 30 people.

Wild T-3's Replaces Parkhursts

Wild T-3 theodolites replaced the Parkhurst, which had been in service for about 25 years, in 1952. Early fears that these small optical reading theodolites were unstable proved unfounded. Some observers found the T-3's stubby gun-sight pointers inadequate for night work and to resolve the problem taped beer can/bottle openers to the top and bottom of the telescopes. Fortunately, this piece of equipment was never in short supply on parties. Later illuminated finders, similar to those found on Parkhursts, were added.

Trivia details aside, economic benefits were accrued by replacing Parkhursts with T-3's. For the most efficient operation the Parkhurst required a 3 person observing unit, an observer, recorder and lightkeeper, who also read the second (B) micrometer. For T-3's the observer makes all the readings and on many occasions the lightkeeper was no longer needed, leaving the slots open for forming more observing units or showing additional lights.

Observation Equations Replace Condition Equations

After the purchase of automatic computing machines in 1948, several test adjustments were made under the overall direction of Charles A. Whitten, Chief, Triangulation Branch, to evaluate the method of observation equations (a k a method of variation of coordinates), both on the ellipsoid and the plane. By 1953, most large and many smaller networks were being adjusted using observation equations and few people had any doubt that the long reign of condition equations was about over.

It was obvious from the beginning that the method lent itself extremely well to automatic computers because the formation of the equations follow identical routines for each specific type of observation, i.e. direction/angle, distance and azimuth. There is no need to study the network to determine the number and kind of condition equations, nor to form them, difficult tasks at anytime. Observation equations require only the identification of the observations and assumed positions be computer recognizable, and similarly for the fixed positions, weights and any conditions placed on the observations.

Until late in the 1960's, it also was necessary to select the normal equations' solution order to minimize the computer space used. After 1970 this was done internally.

About 1957, with the purchase of an IBM 650 electronic computer the method of observation equations officially replaced the method of condition equations, which had been employed for more than 100 years, although the method would continue in use for another 10 years or so. By then, retirement and other forms of attrition had decimated the ranks of those who fully understood condition equations. And, the new generation of geodesists had, at best, only a superficial interest in the method.

In the same time frame, the Cholesky method for solving normal equations was introduced, its fewer steps making it ideal for use with computers. It had not been employed much previously because the fewer steps required didn't translate into significant savings in time and effort for hand computations when compared with the Gauss-Doolittle procedure used since 1878. Furthermore, Cholesky involved square roots that were not easily attained with mechanical calculators then employed. Gauss-Doolittle continued to be used for hand computations and remains a viable method for such calculations.

Computer capacity was limited and only the primary scheme was adjusted simultaneously. Supplemental stations, many positioned by single closed triangles, intersection points and other small jobs were often adjusted by hand, using condition equations.

About 1962, an IBM 1620 electronic computer, with a larger capacity and faster became the work horse of the bureau for a decade before being replaced by a still larger and faster computer. As computer capacity improved, the supplemental and intersection points were adjusted in separate computations and eventually entire networks were adjusted in single weighted computations.

Modern Adjustment Theory Becomes Rule

Early in the change over period, base lines and Laplace azimuths were held fixed in accordance with long practice and considered opinion that these observations were far more accurate than the angle measurements. However, this rationale changed within a few years, as modern adjustment theory that all observations should receive corrections was accepted. By 1965, accuracy estimates were routinely computed as well. Neither practice was really meaningful, because most of the adjustments of the time were badly constrained. From the beginning, accuracy estimates were expressed in terms of probable error and the practice continued until about 1970 when the standard error concept, conforming to modern error theory was adopted.

Data sheets containing all information, other than descriptions, were computer generated by the early 1970's.

Charles A. Whitten - Geodesist

Charles A. Whitten, far more than anyone, brought geodetic computations into the computer age. His successes with punched card computations (1948-57) provided the catalyst and know how needed to adapt the early electronic computers for similar computations. His leadership and interests in adapting the ever improving computers to geodetic needs didn't end with his retirement in 1972. He was often consulted for his expertise in a wide range of geodetic matters as well, continuing to make significant contributions for the next 20 years.

Whitten was a man of many talents, a superb all around geodesist who for more than 40 years served in a variety of positions including Chief of an astronomic party, Chief, Triangulation Branch and Chief Geodesist. A strong advocate for utilizing geodetic surveys to measure crustal motion, Whitten published numerous papers on the results and was convinced such data would one day contribute to the prediction of seismic events.

He was recognized internationally for his work in adjusting the triangulation of western Europe, the basis for the European Datum 1950 and was President, International Association of Geodesy (IAG) 1960-63. Charles A. Whitten died in 1994, at age 84.

Early Geodimeters

The C&GS obtained its first electronic distance measuring instrument (EDMI) a Model 1 Geodimeter in 1953, and the second a Model 2 Geodimeter in 1956. In the period 1953 - 1958, 84 triangulation lines were measured, the first between stations KILIAN and THERESA in eastern Wisconsin, a distance of about 2.25 miles. Four previously taped base lines were included in the total. Each line was observed on at least two separate nights, except for 4 lines in Arizona where only one measurement was made. A measurement consisted of 6 observations, spread over 75 - 90 minutes. The longest line observed was 26 miles, the shortest 0.7 mile and the average 10 miles. All new lines were satisfactorily included in the adjustments of the triangulation.

Both models weighted in excess of 300 lbs. and while good results were obtained from 12 ft. wooden towers, those from a 103 ft. Bilby tower were not, primarily because of stability problems. The problem was resolved in short order. C&GS field personnel were highly skilled in adapting equipment to fit any need, and this situation was no exception.

As a result of these tests, a first-order base line was specified as resulting from 12 complete observations, 6 on one night and 6 on the second.

The most obvious conclusion drawn was the 140 year period of measuring base lines with various mechanical apparatus had ended. Accordingly, the last regular taped base line was measured near Salmon, ID in 1958. Base lines specifically measured to test EDMI were taped at Beltsille, MD in 1964 and near Culpeper, VA (MITCHELL Base) in 1965, with a highly accurate remeasurement in 1967. The first was slightly more than a mile in length, the second about 5.6 miles.

Lasers Give Longer Range

New model Geodimeters were considerably lighter than early versions and weight was no longer a bother, however longer range, especially in daylight and haze was desired. To achieve this, George B. Lesley modified a Model 4D Geodimeter with a 2 milliwatt laser as the light source in 1965. Tests showed conclusively that the desired results were obtained. Ten mile lines were easily measured in bright daylight and at night, a stronger light return at all times. Furthermore, successful measurements under adverse weather conditions were the rule. On the basis of these tests, AGA (Geodimeter Co.) was contracted to convert about 15 Model 4D Geodimeters, then in use in the U.S., to laser instruments.

In the 1970's, Lesley replaced the 2 milliwatt laser in an instrument with a 10 milliwatt one. This extensively modified Model 4D, known as Big Red, routinely measured distances in excess of 50 miles in daylight. AGA and other firms produced laser instruments in the late 1960's.

Tellurometers and Other Microwaves

EDMI using microwave as a measuring source appeared on the scene about 1957. Tellurometer was the first, developed by T.L. Wadley of the South African Research Council. They were lightweight, operated in daylight, had long range capability and less expensive than electro-optical instruments, good features that attracted many surveyors. However, the observations were seriously effected by humidity, especially on longer lines and by a condition known as ground swing. The latter could be corrected by taking certain steps, but was not always recognized.

During the interstate highway program many surveys, including most by the C&GS, were microwave measured traverses and could only be classified as second-order class II (1:20,000) for this reason. Had electro-optical instruments been employed, these surveys would have rated at the very least, the next higher accuracy class.

- Top of Page -