CHAPTER 4 FIRST AIEE PAPER - GRAPHICAL ANALYSIS

From 1943 to 1946, Gordon Walter shared an office with Evans. In a 2004 letter, Walter recounted a pivotal moment in Evans’ intellectual journey:

"Walt and I spent quite a few evenings together in the otherwise deserted Advanced Course office, preparing material for our students while managing our day-to-day jobs on B-20 fire control. One evening, Walt began reading a paper by a Swiss mathematician on a different way to analyze the roots of a polynomial equation. He was intrigued, spent more time thinking about it, and eventually developed an interest in applying it.

We had been exposed to feedback theory in Advanced Course classes, and Nyquist diagrams (Harry Nyquist, on left) were in vogue. Our assignments involved servo systems. For whatever reason, this was when Walt caught the bug that led to root locus analysis and the Spirule. He had the insight, determination, and ability to take the initial concept to a higher theoretical and practical level.

I just happened to be in the same room when the bug first bit him.

The Swiss mathematician was Professor Paul Profos (on right) , whose 1945 article, A New Method for the Treatment of Regulation Problems, published in the Sulzer Technical Review, arrived at the Schenectady library in June 1946. Evans was captivated by it, adding a new dimension to Nyquist’s frequency response method.

Going Home (September 1946)

Shortly thereafter, Walter and Arline Evans made a career-defining decision: Evans would leave General Electric for a faculty position at Washington University in St. Louis. John and Miriam Moore had already decided to leave Schenectady, citing their daughter's health issues caused by the harsh winters. The winter of 1945-46 set a record for consecutive freezing days (36) that still stands.

Many factors likely influenced the Evanses’ decision, with family and familiarity playing a significant role. "You easily make acquaintances among strangers, but genuine friendship requires time to develop," Evans had written to his mother in 1942. That sentiment surely remained true.

But the Evans who returned in 1946 was a changed man. He was now a husband and father. The world had witnessed the defeat of the Axis powers, the death of Franklin Roosevelt, and the dawn of the atomic age. Professionally, he had mastered servomechanisms, taught classes, and discovered Profos’s novel analysis. As an instructor at Washington University, Evans would have the opportunity to further his studies, engage with students and faculty, and—perhaps—publish a paper inspired by Paul Profos’ graphical techniques. 

In September 1946, Walter, Arline, and their son Randy boarded a train at Schenectady’s Union Station and returned home to St. Louis, closing one chapter and opening another.

Upon their returning home to St. Louis from Schenectady, the Evans family moved into the second floor of the house where Arline had grown up—5234 Lotus Avenue—sharing it with her parents, Olinda (née Meyer) and Reinhold Pillisch. Also living there was Arline’s sister, Eleanor, who was unmarried at 32. That status would change a few months later.

Arline’s father was a postal carrier (then commonly referred to as a “mailman”), while her mother was a homemaker. Neither of them ever held a driver’s license, so they relied on public transportation to get around the city, as did Arline. 

Their home on Lotus Avenue, located north of Forest Park, was just a manageable three-mile bicycle ride from Washington University’s hilltop campus on the west side of the park. Whether Walter rode his bike or took the streetcar remains unknown.

Records indicate that Walter taught one graduate course on servomechanisms, using example systems he had worked on in Schenectady. He reported to his former professor, Roy Glasgow, a man known for his humor. Walter recalled one of Glasgow’s aphorisms, such as “The vector sum of all opinions is zero.”

By 1945, Walter had learned the Nyquist and Bode servo diagrams while in General Electric’s Advanced Engineering Program in Schenectady. He had also studied an article titled “A New Method of Regulating System Design” by Dr. Paul Profos, published in the Sulzer Technical Review (Switzerland). Profos’s method involved a curvilinear grid based on a Nyquist curve to determine an eigenvalue at the minus-one point. Walter found this idea compelling and continued exploring it while in St. Louis from 1946 to 1948.

Meanwhile, John R. Moore, who had also returned to Washington University from General Electric, was busy setting up a Mechanics Laboratory. Walter’s mother still lived in Nashville, while his older brother, Cedric, remained in St. Louis, working as a manager at Emerson Electric. His younger brother, Sam, who had laid pipelines across Europe in support of Patton’s 3rd Army before entering the oil business, had moved away. His sister, Alice, had married an Army officer, Duncan Hallock, and relocated as well.

Walter and Arline likely reconnected with old friends from their high school and college years, though specific details aren’t documented. What is certain, however, is that Walter’s fascination with Profos’s ideas about conformal mapping in the complex plane deepened. At Washington University, he found colleagues willing to discuss and critique his ideas. 

Living with Arline’s parents on Lotus Avenue provided the young couple with an opportunity to save money for a future home—and, with their second child on the way, free childcare for Randy.

Walter, always outgoing, surely discussed his ideas with colleagues. Encouraged by their feedback, he wrote to C. F. Wagner, secretary of the Committee on Servomechanisms of the AIEE, just three months into the academic year, inquired about submitting a paper.

December 9, 1946 – Letter to C. F. Wagner 

The purpose of this letter is to learn the procedure for submitting a paper. The proposed title is “Graphical Analysis of Servomechanism Equations.” The idea, believed to be new, is the extension of the Nyquist or Bode diagrams to include the damping constant as well as a real frequency. The usefulness of the method is that the transient response of a system is directly determined rather than merely implied by the steady-state performance... The next meeting listed is not scheduled until June, so I would appreciate knowing if there is an earlier meeting where this paper might be appropriate.