THE WISDOM OF
(A scarce pamphlet published in 1957 By J. R.
Pierce - C- From the Arthur Karp Collection at SMECC)
HARALD TRAP FRIIS
HARALD TRAP FRIIS was born in Naestved, Denmark, February 22, 1893. He
attended the Royal Technical College, Copenhagen, where he received the
degree of Electrical Engineer in 1916 and Doctor of Science in 1938.
He began his engineering career in Denmark in 1916 as assistant to
Professor P. O. Pedersen. From 1917 to 1918 he worked as Technical Advisor
at the Royal Gun Factory in Copenhagen. He received a Fellowship from the
American-Scandinavian Foundation in 1919. That same year he moved to the
United States and, after a period of study at Columbia University, he
joined the Western Electric Company's research department, which was later
to become Bell Laboratories.
During his career with the Bell System, Harald Friis contributed
substantially to almost every aspect of the radio art. His earliest work
was on vacuum tube efficiency. He then worked on ship-to-shore radio
reception. In the early twenties he built the very first field measuring
set, for the 300 to 400 meter range. He also designed the first commercial
double-detection or superheterodyne radio receiver, the Western Electric
In 1923 he spent several months in England setting up receivers for the
long-wave transatlantic telephone link. Later, he worked on short-wave
problems. Here his improvements in receiver noise figure made possible
Jansky's discovery of interstellar radio signals. The observation of
Johnson noise in radio receiver circuits led him to the concept of noise
figure. He engaged in many studies of propagation and of antennas; this
work culminated in the design of the multiple unit steerable antenna (MUSA).
In 1938 Harald Friis entered the field of microwaves. Here he
established accurate measurement techniques which proved invaluable both
in the Bell Laboratories' wartime microwave work and in the work of the
Radiation Laboratory at M.LT. During the war he was engaged in radar work,
particularly in connection with microwave components and antennas. All of
this work was characterized by thorough understanding and accurate
measurements. In the field of antennas, his invention of the simple
rocking horse scanning antenna is particularly noteworthy.
Toward the end of World War II, Harald Friis and his people made
extensive and valuable measurements of microwave propagation. He had the
responsibility for the basic microwave design of the first large-scale
application of microwaves to communication, the TD-X New York-to-Boston
microwave relay system. Thereafter he actively pursued forward-looking
aspects of microwave communication, including an extensive program aimed
at the use of millimeter waves and waveguides for long-route broadband
Although Harald Friis has summed up his work in only a modest number of
publications and patents, none is trivial; each is a milestone in the art.
Harald Friis was made Director of Radio Research in 1945. In 1952 he
became Director of Research in High Frequency and Electronics. He was
awarded the Morris Liebmann Memorial Prize for 1939 by the Institute of
Radio Engineers, and the I.R.E.'s Medal of Honor for 1955. He was awarded
the Danish decoration "Knight of the Order of Dannebrog,"~
presented by King Frederick IX, in 1954, and the Valdemar Poulsen Gold
Medal, presented by the Danish Academy of Technical Sciences, in 1954.
He is a Fellow of the Institute of Radio Engineers and of the American
Institute of Electrical Engineers. His professional activities also
include membership in the American Association for the Advancement of
Science; Danish Engineering Society; American Section, International
Scientific Radio Union; and Danish Academy of Technical Sciences.
December 13, 1957
THE WISDOM OF HARALD FRIIS
WITH NOTES AND COMMENTS
BY HIS DISCIPLE, JOHN PIERCE
Is wisdom wisdom only to the wise? How can we tell wisdom from
unwisdom? If deeds are important, then Harald Friis's deeds show he
must be wise, and many who have worked for him will testify that he
is. He prepared these few notes for a conference at Seaview in
September 1957. I have discussed them with him and added comments,
sometimes in his words and sometimes in mine. What I have added is
NOTES ON RESEARCH JOBS
The following brief notes on research jobs may guide the newcomer in
the field and serve as a reminder to the experienced research worker.
STARTING A JOB
Except for the triggering action of some basic discovery, a job usually
stems from a definite need on a project or a hunch that an advance in the
art is possible. The boss is certainly involved in starting a
comprehensive project. With a smaller job, it is much better for the
initiative to come from the research worker rather than from his boss. It
gives the worker a realization of responsibility, and a successful
solution is more likely.
Basic discoveries, such as those in solid state which led to
transistors, or individual hunches as to possible applications of
physical principles, as in non-reciprocal microwave devices using
Faraday rotation, provide many jobs for the individual worker in a
field. Important research projects, such as ship-to-shore radio,
transatlantic radio, and waveguide communication, were suggested to
Harald Friis by his bosses. Often a man working on a project finds he
has to invent a solution to an unforeseen problem. Thus, while working
on an early unattended receiver, Harald Friis had to invent an automatic
gain control. One often makes interesting discoveries while working on a
project; Harald Friis was the first to observe Johnson noise at radio
frequencies. He did this in the course of making a quieter receiver for
transatlantic radio. This led him to formulate the concept of noise
figure. The receiver was instrumental in the discovery of interstellar
radiation by Karl Jansky.
Where possible, it is desirable to split the job into smaller units. It
is easier for the individual worker to manage and work efficiently on
smaller and well-defined jobs; and remember, that it is always the
aggregate of successfully completed small jobs that produces the
successful big job.
Harald Friis never searched the literature, but then, he has always
been ahead of it. He says, it's good pious advice, though. It is, now that
so many people are doing research in so many places.
A job can be overwhelming when looked at as a whole. But, if it can
be divided into parts and each part made to work with maximum
efficiency, the over-all problem can be solved. Harald Friis says that
when he made radio receivers, he made and tested each part separately,
the oscillator, the mixer, the IF amplifier, and so on. The receiver
always worked when he put it together. Having the responsibility for a
clearly defined part of a project, and one which he can do himself,
helps a man in his work and enables him to take satisfaction in an
accomplishment that is clearly his. Moreover, a boss should not be a
dictator handing out bits of work each morning; he should help a man to
do some clearly defined and worth-while thing.
The research worker should make sure that (a) the job has a well
defined objective, (b) there is a real need or a
possibility for an advance in the art, and (c) the state of the art is
ready for the job. The worth-whileness of the job may be checked by asking
oneself the hypothetical question: "Would I do the job if I had to
pay for it?"
(a) Without an objective, you can't tell when a job is
done. If you go to a man and ask him what he is
doing, you expect an answer.
(b) It would have been possible to transistorize many
microwave systems, but no real need arose. Finally a need arose in
connection with an economical microwave system for
(c) The state of the art was not ready for a waveguide system before
World War II, but it was afterwards.
The remark about paying means if you owned the Bell System, not out
of your own pocket.
The research worker should become obsessed with the job and it
should haunt him day and night if he is the right man for
the job. He may believe that scheduled hours of work limit his freedom
of activity, but actually he has lost his freedom to the job itself.
Excessive amounts of outside activity such as committee work and local
affairs may indicate that the research worker is not in the right
A person has to do some chores for professional
societies, but when a man starts spending a great
deal of time on IRE matters, Harald Friis wonders what's wrong.
A job should be started with a light touch. The research worker
should try different approaches and not collect reams of
data before he has done some preliminary exploring aimed at a simple
solution. He should discuss it with his colleagues and his boss and
consider their suggestions with an open mind. The informal "coffee
table" conference is unbelievably productive in research. He should
also search the literature critically.
Harald Friis never searched the literature, but then,
he has always been ahead of it. He says, it's good pious advice,
though. It is, now that so many people are doing research in so many
WORKING ON A JOB
When working on a job the research worker should not ask others to
solve his problems for him. Certain aspects of the job may require help
from experts in a particular field, but the worker should be the real
leader of his job.
A man has to understand his problem thoroughly before he can ask
intelligent questions about it. If he gets stuck he can seek help, but
he should ask for help, not a solution to his problem. It is the man
responsible for the job who has the motivation to get it done.
The research worker should not be afraid to tackle work in unfamiliar
fields, since he cannot have too broad an experience where his job is
involved. The electrical engineer, for example, should not hesitate to
enter the fields of chemistry, physics, and structural design.
It is easy for a man at the Bell Laboratories to spend lots of
money and suffer endless delays by asking some
other department to do something that he could do quickly and cheaply
(but not without work) himself. One understands best what he does
himself. In the waveguide job, Holmdel men solved problems in
electroforming and in thermosetting plastics. Harald Friis had a lot
of fun in designing a 6o-foot parabolic antenna. He got competent
advice from an expert in structures, but the design was his.
In this age of electronics there is a tendency to use more vacuum
tubes, transistors, and circuits than required for satisfactory
performance. The research worker should guard against this and simplify
his job as much as possible. He should always look for the simplest
experimental and theoretical solutions.
This is extremely important if equipment is to be reliable and
easily maintained. When he built his house, Harald Friis refused to
have an electronic (vacuum tube) thermostat which would hold the
temperature to 1/4 degree. The old-fashioned kind, good to 1/2 degree,
was plenty good enough, and it was simpler and easier to maintain.
It is wasteful to compromise on the best way of doing a research job.
The research worker should always get the special and sometimes expensive
equipment needed for the job.
When Harald Friis had 1200 feet of waveguide put up for experimental
purposes he designed accurate, if expensive, supports and got the most
accurate pipe that could be made. It was scarcely good enough, and
anything less accurate would have been useless.
Harald Friis says you must have a good setup. You must do it the best
way you can think of. You must make good measurements.
It is foolish to measure errors in equipment rather than the
phenomenon you are interested in. On the other hand, one does not use a
sledge hammer to kill a fly.
The research worker is responsible for the work, education, and
development of the technicians assigned to him. Because of this
responsibility, it is annoying to have too much technical help.
Overmanning a job by technicians also has a tendency to foster unnecessary
work. Technical help is required for construction and maintenance of
equipment and for routine measurement, but it is, in general, preferable
to minimize technical help and adopt the "do-it yourself"
Harald Friis says, there should be a rule that whenever something is
made for the first time the engineer should make it himself. Any old
gimmick-any circuit board, anything. When something is duplicated, use a
Technical Assistant. An engineer can tell whether the results make sense
while he is making measurements. A TA may not know. But, J. J. Thomson
was useless with his hands.
The research worker should not start work on a complicated and costly
piece of new equipment in the precision shop before he has had thorough
discussions with the shop foreman on the different ways of doing the job.
Such a screening process is very desirable, since in many laboratories
only a signature on a shop order is required to start a job.
Harald Friis says that one man sent orders to the shop for big pieces
of brass which came back all polished and accurate to a mil, which
wasn't necessary. Sometimes the foreman will know better than an
engineer how to make a thing.
The research worker should stick to his job and get it finished as soon
as possible. Delays caused by shop work are unavoidable nowadays and may
sometimes justify several jobs to be carried on simultaneously. Those jobs
should be closely related, since the worker who
flounders from one thing to another because of too many objectives
generally does not produce.
One can't do everything, but it sometimes takes a long time to find
It is always advisable to work on more than one particular method of
solving a problem. Working on competing methods is stimulating and, like
the farmer, the research worker should not carryall his eggs in one
In the waveguide job, both solid and helical waveguides were
investigated. In the TD-2 job, the triode, the klystron, and the
travelingwave tube were considered.
The research worker should review and re-evaluate his job periodically.
In evaluating a new solution of a problem it is desirable to have an old
solution in mind for comparison purposes. Such a
"standard-of-comparison" solution will naturally change as the
Something new has to be better than something old to be useful. Some
people seem to like things just because they are different. The research
worker should keep his boss informed and make certain that he, the boss,
understands the job thoroughly; the boss must be "on board." He
should never mislead his boss.
A man should know whether or not he is doing what his boss wants. A
man should not go ahead and do something different from what the boss
would do unless he knows he is right. He should have the right to,
though. I don't think that this should be interpreted to mean that the
man should rush to the boss every blue Monday and tell him the job is
going terribly. He should weigh the job and try to give the boss a fair
picture of it. I believe, however, that he should rush to the boss
immediately with all good news. The boss hates to hear such things first
from his boss.
Harald Friis likes the word integrity in connection with research,
but he didn't know quite where or how to get it into his text.
The research worker should record his results in detail in a notebook.
He should also be willing to write an informal quarterly progress report
for the benefit of his associates and boss. The quarterly report may also
provide a valuable background for a later publication.
A patent is important in getting credit for your work and to the Bell
Laboratories. Witnessed notebook entries are invaluable in connection
with patents. Harald Friis said, on the other hand, I never kept a
Contrary to some beliefs, department heads and subdepartment heads
read progress reports; from them they get something of their opinions of
people and jobs. The report is a good historical record, and if well
written it can be incorporated in the text of a published paper. Harald
Friis says that people in my department should write better progress
reports, which he could understand, and that subdepartment heads should
use them to sum up progress on the job from time to time. On the other
hand, writing progress reports can be a chore.
When working on a job there is a tendency for the creative research job
to turn into the "bread-and-butter" type job that occupies more
than 50 per cent of the time of the technical people in a research
laboratory. The bread-and-butter job requires competent technical help,
but the highly creative worker can now relax and explore new fields.
As a job gets well in hand it can be taken over by less creative
people. This takes a bad burden off the creative man.
By-products may change the course of a job. Such changes are natural
and permissible in research.
Despite an investment in equipment and an identification with an
approach, one should be willing to throw the old away if some other
approach is clearly better. Here is one place where integrity is
important. Harald Friis had a wooden tower built. He watched it go up.
He took a good look at it when it was done and said, that's not the way
to build a tower.
One shouldn't be so obsessed with matters at hand as to overlook more
important things. If I hadn't been so hypnotized with the idea of an
amplifier I would have seen the full importance of backward wave
oscillations when Millman produced them. Sometimes byproducts are more
important than the job itself.
The research worker should be happy if he was brought a job to a stage
where other workers make use of his results. If his job is so successful
that great expansion of effort is indicated, he should feel complimented
rather than disappointed that he is no longer in complete control of the
Some people think they would like to have a sort of patent on a field
of work. It is, however, a real accomplishment to sell your work to
Excepting long-range jobs, such as advancing the arts of radio,
electronics, or waveguides, a job may be stopped before it is finished.
This is most difficult, since research jobs have a tendency to go on and
on; further improvement is always around the corner. But, if the
importance of the job has decreased and if favorable results are too
meager, then the job should be stopped. When in doubt, it is better to err
by stopping a job, since many jobs may benefit by a period of rest and may
readily be resumed if justified later.
In the thirties Harald Friis worked on short waves, rhombic antennas,
and MUSA. Ralph Bown told him to stop all this work and go into
microwaves. He says, it took me a week to get over this.
FINISHING A JOB
In finishing a component or systems type of job, the research worker
must have carried his work to the stage where he can demonstrate his
results by means of a finished research model or system. He should publish
his results in a logical and concise manner, memorandum, paper, or patent,
and he should see his job to development. He should then be willing to
drop the job and think about the future.
If you insist on perfecting what you are working on, you will never
have time to do anything new.
It is important to the morale of the worker and his laboratory that his
research job be evaluated correctly. For example, does the job represent a
real milestone in the advance of the art or is it just a worth-while
improvement? It is unfortunate that the bosses in large laboratories often
do not appreciate the value of the work of their people, or at least they
do not show that they do.
These are hard words, but there is some truth in them. Bosses should
try to get behind public acclaim and popular ideas. Big bosses should
show appreciation of fundamental advances. No one is above taking
pleasure in appreciation and praise. If they don't get praise for the
right things they won't learn right from wrong. It is also important, as
noted earlier, for people down the line to tell bosses what is
The research worker should realize that he is not to blame if Nature
has not permitted favorable results. He is responsible for, and should be
satisfied with, accurate quantitative results.
I have seen a good man work for years on something that proved
impossible. Knowing what can't be done can be very valuable, as we see
in the case of perpetual motion. Harald Friis says that MUSA didn't give
as good results as people had expected.
The research worker must accept gracefully the possibility that, ahead
of him, someone in another laboratory has found a solution to his problem.
With so much effort along broad fronts nowadays, all research workers are
exposed to such disappointments.
No one gets anything but patents in struggles for priority. People
should take such things gracefully. There is nothing more annoying or
pitiable than a man emotionally involved in showing that he was first.
Moreover, if you are interested in getting the job done, the origin of
the idea is less important than the fact that it is the right one.
The research worker should remember that some credit for a successful
job, perhaps more than 50 per cent, belongs to the laboratory employing
him. The laboratory supplied a large amount of know-how and gave him many
aids and inducements to carryon his work.
Harald Friis says that the percentage varies with the individual, and
that in many cases the Laboratories deserves more than 50 per cent
credit. He says, the fact that I was planted in the Labs, and all the
background, that was everything, John.
Antennas: Theory and Practice (with S. A. Schelkunoff). New York: John
Wiley and Sons, 1952.
Radar Systems and Components, Chapter on "Radar Antennas"
(with W. D. Lewis), pp. 763-858. New York: D. Van Nostrand Company, 1949.
"The Vacuum Tube as a Generator of Alternating-Current Power"
(with John H. Morecroft), Proceedings of the AlEE, Vol. 38, No. 10,
"Static Interference as a Function of Wavelength" (with L. J.
Sivian), Wireless World and Radio Review, Vol. 10, No. 10, June 3, 1922.
"Radio Transmission Measurements" (with R. Bown and C. R.
Englund), Proceedings of the IRE, April 1923, pp. 115-152.
"High-Frequency Amplifiers" (with A. G. Jensen), Bell System
Technical Journal, April 1924, pp. 181-205.
" A New Directional Receiving System," Proceedings of the
IRE, December 1925, pp. 685-707.
"A Static Recorder," Bell System Technical Journal, April
1926, pp. 282-291.
"A Radio Field-Strength Measuring System" (with E. Bruce),
Proceedings of the IRE, August 1926, pp. 507-519,
"Methods for the Measurement of Radio Field Strength" (with
C. R. Englund), Transactions of the AIEE, 1927, pp. 492-497.
"Direction of Propagation and Fading of Short Waves,"
Proceedings of the IRE, May 1928, pp. 658-665.
"Some Effects of Topography and Ground on Short-Wave
Reception," Proceedings of the IRE, April 1932, pp. 699-721.
"Determination of Direction of Arrival of Short Radio Waves"
(with C. B. Feldman and W. M. Sharpless), Proceedings of the IRE, January
1934, pp. 47-78.
"A Multiple-Unit Steerable Antenna for Short-Wave Reception"
(with C. B. Feldman), Proceedings of the IRE, July 1937, pp.
"Noise Figures of Radio Receivers" Proceedings of the IRE,
Vol. 32, July 1944, PP·419-422.
"Note on a Simple Transmission Formula,"' Proceedings of the
IRE, Vol. 34, May 1946, pp. 254-256.
"Radar Reflections from the Lower Atmosphere," Proceedings of
the IRE (Letter to the Editor), Vol. 35, No.5, May 1947.
"Microwave Repeater Research," Bell System Technical Journal,
Vol. 27, April 1948, pp. 183-246.
"A Reflection Theory for Propagation Beyond the Horizon"
(with A. B. Crawford and D.C. Hogg), Bell System Technical Journal, Vol.
36, May 1957, pp. 627-644.
and L. C. Tillotson)
6-1-26 Duplex Carrier Wave
Transmitting and Receiving Systems
11-1-27 Radio Signaling Apparatus
7-3-28 Transmission Regulation
7-24-28 Signal Reception
9-3-29 Radio Receiving Circuits
(with A. G. Jensen)
6-10-30 Radio Receiving Systems
3-10-31 High Frequency Oscillation
5-5-31 Wave Amplifying (with
2-16-32 Radio Receiving Circuits
5-24-32 Wave Transmission Circuits (with
A. G. Jensen)
7-12-32 Electric Wave Translating Systems
7-4-33 Radio Direction Finders
(with E. Bruce)
2-13-34 Antenna Counterpoise Systems
9-3-35 Translating Circuit
5-19-36 Radio Systems
3-25-41 Directional Antennas (with A. C.
6-17-41 Radio System (with C. B. H.
6-11-46 Coupling System
8-27-46 Oscillation Generator
3-4-47 Horn Antenna Systems (with A.
4-29-47 Branching Circuits
11-20-51 Microwave Hybrid Junction (with W. D.
3-3-53 Waveguide Joint (with
A. C. Beck)
12-11-56 Transmitting and Receiving Circuits for
Wave Transmission Systems