Apuntes de la HP
C++
The C++
Programming Language,Bjarne Stroustrup http://www.research.att.com/~bs/3rd.html
Sentencias de Control en C++, http://www1.ceit.es/Asignaturas/Informat2/Apuntes/Sentencias.pdf
Opening Black’s Box
Rethinking Feedback’s Myth of Origin
DAV I D A . M I N D E L L
The specific triumph of the technical imagination
rested on the ability to dissociate
lifting power from the arm and create a crane: to
dissociate work from
the action of men and animals and create the
water-mill: to dissociate light
from the combustion of wood and oil and create the
electric lamp.
—Lewis Mumford
The engineer who embarks on the design of a
feedback amplifier must be a
creature of mixed emotions.
—Hendrik Bode
Like any modern episteme worthy of the name, the
theory of feedback has
a myth of origin. On a sunny August morning in
1927, Harold Black, a
twenty-nine-year-old systems engineer, rode the
Lackawanna ferry to work
at the Bell Telephone Laboratories.Many Bell
engineers lived in New Jersey,
and on the early morning ferry rides across the
Hudson to the Manhattan
laboratories they frequently gathered on the
forward deck. This morning
Black stood alone, staring at the Statue of
Liberty, and had an epiphany: “I
suddenly realized that if I fed the amplifier
output back to the input, in
reverse phase, and kept the device from oscillating
(singing, as we called it
then), I would have exactly what I wanted: a means
of canceling out the distortion
in the output.”1 As it happened,
the New York Times that day con-
Dr. Mindell is Dibner Associate Professor of the
History of Engineering and Manufacturing
at the Massachusetts Institute of Technology. He
thanks Annette Bitsch, John
Staudenmaier, and anonymous reviewers for helpful
feedback to stabilize this paper, and
Sheldon Hochhieser of the AT&T archives and
Ronald Kline for their help with documentary
research.
©2000 by the Society for the History of Technology.
All rights reserved.
0040-165X/00/4103-0001$8.00
1. Harold S. Black, “Inventing the Negative
Feedback Amplifier,” IEEE Spectrum 14
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
406
tained a blank page, and Black sketched
his idea, “a simple canonical diagram
of a negative feedback amplifier plus the
equations for the amplification
with feedback.” He rushed into work,
asked a technician to wire up a
prototype, and gave birth to a
foundational circuit of modern electronics.
This story has become enshrined as one
of the central “flashes of insight”
in electrical engineering in this
century, periodically retold as an inspiration
for engineers.2 A common textbook on control engineering reprints
the story of Black’s vision verbatim in
the first chapter.3
At Bell Laboratories from 1927 to 1940,
the legend goes, Black, Harry
Nyquist, and Hendrik Bode laid the
foundations of feedback control that
engineers then applied to all types of
closed-loop systems, from servomechanisms
to thermostats, fire control systems to
automatic computers.4
More than other contemporary narratives
of control systems such as automatic
pilots or servomechanisms, this story
of feedback earned a place in
engineering legend and college
textbooks. It produced design methods and
graphical techniques that carry their
author’s names (the Bode plot, the
Nyquist diagram) and earned telephone
engineering a claim to priority in
feedback history. Feedback theory,
moreover, formed the basis of cybernetics,
systems theory, and a host of other
post–World War II information sciences,
so Black’s invention is hailed as a
foundation of the information age.
Feedback is indeed a fundamental
concept in twentieth-century technology,
and the Bell Labs feedback theorists
did lay critical foundations for
it. But the origin myth effaces its
sources. It skips over the inventors themselves
and the ways in which their backgrounds
and prior experience influ-
(December 1977): 54–60. George
Stibitz’s memoir describes the early morning ferry
rides; “The Zeroth Generation,”
manuscript, 1993, Stibitz Papers, Dartmouth College,
54. For the quotations from Mumford and
Bode, see Lewis Mumford, Technics and
Civilization (New York, 1934), 33; Hendrik Bode,
“Relations Between Attenuation and
Phase in Feedback Amplifier Design,” Bell System Technical Journal 19 (July 1940):
421–54.
2. For other accounts of Black’s
invention, see Hendrik Bode, “Feedback: The
History of an Idea,” Proceedings of the Symposium on Active
Networks and Feedback Systems
(Brooklyn, 1960), reprinted in Selected Papers on Mathematical Trends
in Control
Theory, ed. Richard Bellman (New York, 1964);
M. J. Kelley, “Career of the 1957 Lamme
Medalist Harold S. Black,” Electrical Engineering 77 (1958): 720–22; Prescott C. Mabon,
Mission Communications: The Story of
Bell Laboratories (Murray
Hill, N.J., 1975), 39–40.
Among historians’ accounts the most
thorough is Stuart Bennett, A History of Control
Engineering, 1930–1955 (London, 1993), chap. 3, “The
Electronic Negative Feedback
Amplifier.” See also E. F. O’Neill,
ed., A
History of Science and Engineering in the Bell
System: Transmission Technology
(1925–1975) (Murray
Hill, N.J., 1985), chap. 4, “Negative
Feedback”; Ronald Kline,“Harold Black
and the Negative-Feedback Amplifier,” IEEE
Control Systems (August 1993): 82–85; and a short film,
Communications Milestone:
Negative Feedback (Bell Telephone Laboratories, 1977).
3. Richard C. Dorf, Modern Control Systems, 5th ed. (Reading, Mass., 1995).
4. Hendrik W. Bode, Synergy: Technical Integration and
Technological Innovation in
the Bell System (Murray Hill, N.J., 1971), 138–40.
enced their work. It reveals little
about the concrete problems these men
worked on when they produced their
solutions. The story also removes
feedback theory from its engineering
culture, that of the telephone network
between the world wars. Black’s version
also does not account for the relationship
of his feedback amplifiers to prior
traditions of governors and selfregulating
machinery.
Thus a reexamination of the sources is
in order, retelling Black’s legend
not as a heroic tale but as the story
of an engineer solving the technical
problems of a particular place and time
and trying to convince others to
support his solutions. As it turns out,
Black did not understand as much
about feedback as he later recalled. To
make his idea credible, he needed
Nyquist’s reformulation of the problem
of stability and Bode’s analysis outlining
the tight constraints that a feedback
amplifier had to meet. He also
needed the Bell System.
Negative-feedback amplifiers emerged from efforts
to extend the telephone network across
the continent, to increase the network’s
carrying capacity, and to make it work
predictably in the face of
changes in season, weather, and
landscape—from the context, that is, of
building a large technical system and
operating it over a diverse and
extended geography. Black, Nyquist, and
Bode worked within a company
that sought to translate ever more of
the world into transmissible messages.
This translation required, among other
things, ever closer couplings of
human and mechanical elements through
the medium of sound, couplings
that left a discernible mark on
feedback theory. For telephone engineers,
the network listened, and it spoke.
In 1934, the same year that Black
published his amplifier, Lewis Mumford,
in Technics and Civilization, noted technology’s ability to
abstract the
world. “Men became powerful to the
extent that they neglected the real
world of wheat and wool, food and
clothes,” he wrote, “and centered their
attention on the purely quantitative
representation of it in tokens and
symbols.”5 In light of Mumford’s observation, a retelling of
Black’s story
has greater significance than a simple
corrective to the origin myth, for it
concerns the historical emergence of
electrical signals as representations of
the world, the technologies developed
to manipulate and transmit them,
and the economic and organizational
conditions that made those technologies
possible. Black, Nyquist, and Bode
contributed to an understanding
of telephony as the transmission of
abstract signals, separate from the
electric waves that carried them. The
AT&T engineers’ increasing facility
with creating, manipulating, and
switching such signals prompted them to
rethink the network not simply as a
passive medium but as an active
machine. Then the Bell System became
not merely a set of voice channels
but a generalized system, capable of
carrying any signal as a new currency:
information.
MINDELLK|KRethinking
Feedback’s Myth of Origin
407
5. Mumford, 25.
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
408
Network Geography
The Bell System of 1900 was an
engineer’s dream: geographically
expansive, reaching into all types of
difficult terrain and climates, and yet
always in control, tied to the central
office. Still, in the first decade of the
century American Telephone and
Telegraph (AT&T) did not yet have its
later hegemony. The company controlled
only about half the telephones in
the country, and long distance was the
key to expanding that share.
Originating in New York, the Bell
System followed its own frontier on a
western expansion.6 From the turn of the century until the 1930s,
AT&T
expressed its technical milestones in
geographical terms: the New York/
Chicago line stood for
carrier-frequency transmission; the New York/San
Francisco transcontinental line stood
for vacuum-tube repeaters; the
Morristown trial simulated the entire
country and represented the negative-
feedback amplifier.“People assimilated
telephony into their minds as if
into their bodies,” writes telephone
historian John Brooks, “as if it were the
result of a new step in human evolution
that increased the range of their
voices to the limits of the national
map.”7
THE PASSIVE NETWORK
Despite these ambitions, at the turn of
the century the telephone network
remained a passive device, as it had
been since Bell’s invention.
Carbon microphones added energy from a
battery to the weak acoustic signal
from a speaker’s voice, but once the
wave entered the line it traveled to
the receiver without further
amplification, going the full distance on its
original strength. In fact, impedance
in the wire imposed considerable
losses, known as “attenuation.”Around
1900 the telephone network ran up
against the limits of transmission,
both in extension, which determined the
furthest distance a signal could
travel, and in economy, which determined
the cost of sending a signal over
shorter distances.
Weather exacerbated the problem. The
standard method of transmission,
even for long distances, was “open
wire,” which meant each circuit literally
had its own wire, separated from others
by a few inches of space. This
separation minimized cross talk, where
one conversation leaked to an adjacent
wire, and also kept attenuation losses
to a minimum. Telephone poles
6. For the general history of the Bell
System, see John Brooks, Telephone: The First
One Hundred Years (New York, 1975); Thomas Shaw, “The
Conquest of Distance by
Wire Telephony,” Bell System Technical Journal 23 (October 1944); Leonard Reich,
“Industrial Research and the Pursuit of
Corporate Security: The Early Years of Bell
Labs,” Business History Review 54 (winter 1980): 511. See also Leonard
Reich, The
Making of American Industrial Research:
Science and Business at GE and Bell, 1876–1926
(New York, 1985), chaps. 7–8. For
another interpretation of the semiotics of telephony,
see Avital Ronell, The Telephone Book: Technology,
Schizophrenia, Electric Speech
(Lincoln, Neb., 1991).
7. Brooks, 142.
MINDELLK|KRethinking
Feedback’s Myth of Origin
409
with dozens of wires (familiar in
turn-of-the-century urban scenes) distinguished
this technology. In addition to
cluttering the landscape, the lines
were particularly vulnerable to snow
and ice storms. Cables, an alternative to
open wire, collected numerous small
wires together into a thick bundle. They
could be buried underground, which made
them immune to weather and
cheaper to install. But because the
wires were of small diameter and packed
tightly together, cables had higher
losses than open wire, twenty to thirty
times more signal attenuation, so they
lowered the limits of transmission.
To push these limits, Michael Pupin of
Columbia University and
George Campbell of Western Electric,
working simultaneously, developed
the loading coil. By adding inductance
at intervals along the wire, loading
coils could decrease signal loss by a
factor of three or four, and thus increase
the maximum transmission distance
proportionally.8 Commercial installation
began in 1904, and loading coils
rapidly proliferated through the network,
especially on cabled routes.9 Still, the loading coil remained passive—
it facilitated the propagation of the
wave down the line but added no
additional energy.
THE TRANSCONTINENTAL LINE: GEOGRAPHY
AND STANDARDIZATION
Not only technical innovations but also
organization and policy supported
the network’s expansion. John J. Carty,
chief engineer of the Bell
System in 1907, had a clear vision of
the social role of the telephone network
as “society’s nervous system.” He and
his engineers vigorously pursued
the goals of AT&T President
Theodore Vail’s famous motto: “One policy,
one system, and universal service.”
Carty strongly supported science
within the company. He had a vision of
industrial research that translated
corporate goals into technical problems
to be solved in the laboratory
(sometimes as much for protection
against competition as for advancement).
10 One of Carty’s longtime associates recalled him as
a system-
8. James E. Brittain, “The Introduction
of the Loading Coil: George A. Campbell and
Michael I. Pupin,” Technology and Culture 11 (1970): 36–57. See also the
discussion of
Brittain’s article by Lloyd
Espenschied, Joseph Gray Jackson, and John G. Brainerd,
Technology and Culture 11 (1970): 596–603. Neal Wasserman, From Invention to
Innovation: Long-Distance Telephone
Transmission at the Turn of the Century (Baltimore,
1985).
9. M. D. Fagan, ed., A History of Engineering and Science in
the Bell System: The Early
Years (1875–1925) (Murray Hill, N.J., 1975), 241–52.
10. Ibid., 32–35, 44. Ironically, in a
consolidation of research, Carty closed Western
Electric’s Boston engineering
department, which had been investigating Lee De Forest’s
audion for use as an amplifier. Hugh
Aitken argues that the closing of the lab may have
cost the company several years toward
making a practicable telephone amplifier. A proposed
contract with Reginald Fessenden for
radio technology also became a casualty of
Vail’s consolidation. “What slipped
through the Telephone Company’s fingers, in short,
was a unique opportunity to come to
grips with electronic technology,” Aitken argues,
countering other historians (Hoddeson
and Reich) who view the move to a single department
in New York as progress toward industrial
research; see Hugh Aitken, The
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
410
builder in the Hughesian sense: “He
recognized the interrelationship in the
telephone business of operating
methods, design of the plant, and the rate
structure. . . . He had in mind that
all of these factors must be considered
in relations to one another.”11 And on all of these factors, Carty believed,
science could be brought to bear.
And science he needed. By 1911, the
state of the transmission art had
hit its practical limit: “loaded” lines
reached the 2,100 miles between New
York and Denver, but the attenuation
and distortion so mangled voice signals
they were barely understandable. Yet in
1909 AT&T’s technical management
initiated a project to extend the Denver
line to California, completing
a transcontinental line. This
geographical problem had a technical
core. Bridging the distances required
an amplifier or “repeater,” an active
device that added energy to the signal,
unlike loading coils, which merely
stemmed its decay.12 To solve this problem, in 1911 Carty organized a
special
Research Branch of the Western Electric
Engineering Department, with
E. H. Colpitts as its head.13
The solution to this problem of
long-distance transmission emerged
from a new alliance of corporate
interests and the latest academic science.
Carty gave technical responsibility for
the transcontinental line to a young
physicist, Frank Baldwin Jewett. Jewett
came to Western Electric in 1904
from a stint as an instructor in
electrical engineering at MIT.He had earned
Continuous Wave: Technology and
American Radio, 1900–1932 (Princeton, 1985), 75–78.
Lillan Hoddeson, in “The Emergence of
Basic Research in the Bell Telephone System,
1875–1915,” Technology and Culture 22 (1981): 530, notes that the term
“fundamental
research” began to appear in the
company’s rhetoric about 1907, a point echoed in Horace
Coon, American Tel and Tel: The Story of a Great Monopoly
(New York, 1939), 197.
See also
Reich, “Industrial Research” and The Making of American Industrial
Research,
for the
defensive stance of early industrial
research.
11. Bancroft Gherardi, “The Dean of
Telephone Engineers,” Bell
Laboratories Record
9 (September 1930).
12. Mechanical telephone repeaters,
logical extensions of simple and common telegraph
repeaters, had existed for some time.
These devices coupled acoustic energy from
a speaker into a microphone, amplified
the signal, and retransmitted it. This approach
amounted to connecting two telephone
circuits end to end, and numerous such devices
were patented before 1900. More elegant
solutions used the same principle but combined
the elements into a single unit.
Because of inertia, the mechanical coupling lagged
the electrical signal and the output
was not very linear with input, which meant that
mechanical repeaters introduced
significant distortion. No more than a few could be
connected in series, and the delicate
devices proved especially sensitive to temperature
variations. Developing a repeater had a
strategic dimension as well: the rapid rise of new
wireless communications seemed a threat
to wired communication, and repeaters would
give the company the opportunity to
control radio technology, which required similar
types of amplifiers. Shaw (n. 6 above)
reprints Carty’s original proposal for the transcontinental
line.
13. The organization charts of the
AT&T/Western Electric Engineering departments
in 1905, 1907, 1909, 1911, 1915, and
1925 are reprinted in Shaw (n. 6 above), 400–406,
and Fagan, 43–55.
MINDELLK|KRethinking
Feedback’s Myth of Origin
411
his doctorate in physics at the
University of Chicago, where he worked
under Albert A. Michelson and became
friendly with Robert Millikan. In
1910, faced with the problem of making
repeaters for the transcontinental
line, Jewett imagined that a solution,
“in order to follow all of the minute
modulations of the human voice, must be
practically inertialess.”14
Mechanical repeaters had existed for
some time, but they were impracticable
because the inertia of their elements
introduced significant distortion.
Jewett thought the secret to
“inertialess,” and hence high-quality, repeaters
lay in the electron physics he had
studied at Chicago. At his request,
Millikan sent several recent Ph.D.’s to
AT&T to work on the project, and
they formed an important axis of the
company’s research for years to come.
In the ensuing decades Jewett would
become an important figure in
American science, but within AT&T
his name was intimately associated
with long distance transmission.When he
retired in 1944, Bell Laboratories
published an “implicitly biographical”
tribute: not a description of the
man’s life, but a detailed technical
history of the transcontinental line.15
After Jewett, Harold D. Arnold was the
first of the Chicago group to
arrive at AT&T, where he joined
Colpitts’s new Research Branch. Arnold,
with fellow Millikan disciple H. J. van
der Bijl, analyzed electron behavior
within De Forest’s audion tubes,
characterized the tubes’ behavior as circuit
elements, and engineered them for mass,
interchangeable production. By
1913, Arnold’s “high vacuum thermionic
tube,” later known simply as the
vacuum tube, could amplify signals in
telephone repeaters.16 This electronic
repeater made possible the transcontinental
line, which opened at the Pan
American Exposition in San Francisco in
1915 with great fanfare. From the
east coast, Alexander Graham Bell
repeated his famous first conversation
with Thomas Watson, now in California.
Vail and President Woodrow
Wilson both chimed in as well. The line
consisted of 130,000 poles, more
than 99 percent on open wire (the few
cables forded streams and rivers). It
had loading coils every eight miles and
eight vacuum-tube repeaters ampli-
14. Jewett to Millikan, quoted in Fagan,
258. Jewett and Millikan had boarded
together at Chicago, and Jewett was the
best man at Millikan’s wedding. Robert A.
Millikan, The Autobiography of Robert A. Millikan
(New York, 1950), 52–53.
Millikan
recounts the story of Jewett’s approach
to him, 116–17. Millikan remained a consultant
in long-distance telephony, and his
testimony helped settle the protracted suit between
General Electric and AT&T over the
vacuum tube, 120–22.
15. Shaw, 533. Bruno Latour uses
Jewett’s appropriation of the electron as an example
of “machines” as abstract apparatuses
for tying together interested groups; Science in
Action (Cambridge, 1987), 125–26.
16. Shaw, 375, 379–82. Hugh Aitken
argues that Arnold simply had a fundamentally
different vision of the audion’s
potential than did De Forest. “Arnold . . . saw in it . . .
something its inventor did not see: the
possibility of making it into a high-vacuum
device, operating by pure electron
emission,”whereas De Forest saw it as a gas-discharge
device. Still, in Aitken’s view, the
distance between telephony and wireless delayed the
Bell system’s adoption of the audion
for a number of years. Aitken (n. 10 above), 546.
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
412
fying the signal in both directions. Still,
calling across country was far from
routine; a three-minute call cost more
than twenty dollars, and delivered
only a third of the bandwidth of
standard lines, which meant greatly
reduced quality.17 Its scratchy tone notwithstanding, the transcontinental
line brought the entire country within
the scope of Vail’s unifying vision.
Amid the fanfare, however, the
transcontinental line also marked a lessnoted
but equally critical technical and
conceptual shift: the network
became a machine. No longer was the
network a passive device; with
repeater amplifiers, the network
actively added energy along the route, a
significant change because it
effectively decoupled the wave that represented
the conversation from its physical
embodiment in the cable. Electricity
was no longer the conversation itself,
but “useful only as a means of
transmitting intelligible sounds . . .
[with] no appreciable value purely from
the power standpoint.”18 In other words, a working amplifier could renew
the signal at any point, and hence
maintain it through complicated manipulations,
making possible long strings of
filters, modulators, and transmission
lines. Electricity in the wires became
merely a carrier of messages, not
a source of power, and hence opened the
door to new ways of thinking
about communications.
Standardization accompanied the
conceptual shift. Once voices became
signals, they could be measured and
specified. No longer did the system
merely deliver conversations according
to some vague notion of clarity.
Now the telephone company delivered
products: signals within a specific
frequency range, at a specified volume,
and with a specified amount of
noise. This transformation required
standard measures: the “mile of standard
cable,” for example, became the
“transmission unit,” renamed the
“bell,” and eventually standardized in
the “decibel,” smaller by a factor of
ten (and still today the standard
measure of attenuation). Noise itself
became a measurable quantity (based on
thermodynamics), and the limiting
factor in quality.19 The message was no longer the medium, now it was
a signal that could be understood and
manipulated on its own terms, independent
of its physical embodiment.
17. E. H. Colpitts,“Dr.H.D.Arnold,” Bell Laboratories Record 6 (June 1928): 411–13.
Actually,mechanical repeaters initially
carried the transcontinental line but were quickly
replaced with electronic ones.
Gradually, more repeaters were added and the number of
loading coils reduced; the coils
reduced the bandwidth of transmission, and also reduced
the speed of signal propagation, which
led to problems with echoes. Shaw, 389–92, provides
a detailed technical description of the
transcontinental line. The line was not permanent
but rather was “built up by
switches”when needed, as was the New York/Denver
line. Fagan (n. 9 above), 263–64.
18. H. H. Nance and O. B. Jacobs,
“Transmission Features of Transcontinental Telephony,”
Journal of the American Institute of
Electrical Engineers 45
(1926): 1062.
19. W. H. Martin, “Transmitted
Frequency Range for Telephone Message Circuits,”
Bell System Technical Journal 9 (July 1930): 483–86, and “The
Transmission Unit and
Telephone Transmission Reference
Systems,” Bell
System Technical Journal 3 (July 1924):
MINDELLK|KRethinking
Feedback’s Myth of Origin
413
A Signal Organization
The success of the transcontinental
line proved to Carty and AT&T the
value of Jewett’s alliance of physics,
electronics, and telephone engineering.
20 Duplicating this success in other arenas, however,
would require an
organizational solidity as well. On 1
January 1925, the AT&T and Western
Electric engineering departments
combined to form the Bell Telephone
Laboratories Incorporated (BTL). The
new entity was responsible to AT&T
for fundamental research and to Western
Electric for the products of
research, and the two companies funded
it accordingly. Located at 463 West
Street in Manhattan, the lab had
thirty-six hundred employees, including
two thousand scientists and engineers. Carty
served as chairman of the
board, which also included vice
presidents of Western Electric and AT&T.
Frank Jewett became president, and
Harold Arnold director of research.
While an important milestone for
corporate research, it is easy to overestimate
the importance of the foundation of the
laboratory itself. The new
organization resembled the old Western
Electric engineering department
with only moderate changes.21 Research conducted at Western Electric carried
on largely unaltered, as did the
careers of the engineers. Indeed, it
would be inaccurate to characterize all
of BTL’s work as industrial research
addressing fundamental scientific
problems. Most of the staff of BTL,
including Harold Black, engaged in the
creative, if routine, work of designing
telephone equipment and making it work.
Despite the system-oriented
organization, no group within BTL did
“system engineering” in the
post–World War II sense. The systems
development department, to which
Black belonged, did not formulate an
abstract vision of the system overall,
but in fact designed the actual
circuits for the network, including equipment
structures, office layouts, and the
electric power systems required to
run the equipment.22
Only the research department performed
fundamental industrial
research in the classical sense. Headed
by Harold Arnold and comprising
five hundred people, its mission was
“to find and formulate broadly the
400–408. R. V. L. Hartley, “TU Becomes
‘Decibel,’” Bell
Laboratories Record 7
(December
1928): 137–39. J. B. Johnson, “Thermal
Agitation of Electricity in Conductors,” and H.
Nyquist, “Thermal Agitation of Electric
Charge in Conductors,” Physical
Review 32
(1928): 97–113.
20. The transcontinental line so
solidified the alliance technically that loading coils
were gradually removed from the
network. The transcontinental line was fully unloaded
in 1920, more than tripling the
velocity of transmission, which reduced echo effects and
improved the “sense of nearness” of the
speakers. Shaw (n. 6 above), 396.
21. Fagan, 54–55, compares BTL with the
old AT&T and Western Electric engineering
organizations. Also see the
organization charts in Shaw, 406, for its similarity to the
initial BTL organization outlined
below.
22. Paul B. Findley, “The Systems
Development Department,” Bell Laboratories
Record 2 (April 1926): 69–73.
laws of nature, and to be concerned
with apparatus only insofar as it serves
to determine these laws or to
illustrate their application in the service of the
Bell System.”Research covered nine main
areas: speech, hearing, conversion
of energy between acoustic and electric
systems (speakers and microphones),
electric transmission of intelligence,
magnetism, electronic physics,
electromagnetic radiation, optics, and
chemistry.23 Yet even within the
Bell System, the research department
did not have a monopoly on fundamental
exploration, because the development
and research (D&R) department
of AT&T, with a similar charter and
eleven hundred engineers and
scientists, remained separate from BTL
for the labs’ first ten years. The negative-
feedback amplifier emerged from
interactions, and even conflicts,
between the concrete, technical culture
of systems development and the
more theoretical research world growing
within BTL.
The Technical Charge of Bell Labs
After the New York to San Francisco
line in 1915, wires couldn’t go
much further (crossing the oceans was
considered a problem for radio).
But it was one thing to span the
continent and quite another to offer highcapacity,
economical service over that distance.
Just keeping up with growing
demand proved a constant problem: the
Bell System added eight hundred
thousand new subscribers in 1925 alone.
Such expansion entailed
planning and forecasting future
requirements based on the rate of growth
and detailed cost analysis to determine
when new technologies were
required.24 Engineering studies evaluated a series of tradeoffs
between the
diameter of the wire, the number of
repeaters, the cost of the terminal
equipment, and the number of available
channels. Increasing the capacity
over existing long-distance routes, and
thereby cutting costs, began to drive
transmission development at Bell
Laboratories.
Bell Labs engineers thus turned their
attention to putting more conversations
onto a single line. The most promising
method, carrier multiplexing,
modulated several voice signals onto
high-frequency carrier signals. If
these modulations occur in distinct
frequency bands they can all travel over
the same line, in much the same way
that separate radio stations occupy the
single electromagnetic spectrum
(indeed, the technique became known as
“wired wireless”).25 At the receiving end, a wave filter separates out
the
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
414
23. Paul B. Findley, “The Research
Department,” Bell
Laboratories Record 2
(June
1926): 164–70.
24. H. P. Charlesworth, “General
Engineering Problems of the Bell System,” Bell
System Technical Journal 4 (October 1925), 515–41.
25. John Stone Stone, “The Practical
Aspects of the Propagation of High Frequency
Electric Waves Along Wires,” Journal of the Franklin Institute 174 (October 1912),
described high-frequency multiplex
telephony as “identical with that of the new continuous
wave train” radio, and included the
Alexanderson alternator as an element of a tele-
MINDELLK|KRethinking
Feedback’s Myth of Origin
415
voice channel (figs. 1 and 2). The idea
had been around for a long time:
both Elisha Gray and Alexander Graham
Bell had investigated carrier techniques
in their telephone research.26 But vacuum tubes made carrier
telephony practicable by allowing
signals to be cleanly modulated, filtered,
and amplified. The first commercial carrier
system, type A, was installed in
1918, putting four two-way channels on
open-wire pairs.27 Still, carrier had
its problems: because of the high
frequencies, carrier signals faced greater
attenuation than traditional voice-band
signals and hence required more
repeaters.
Another means of increasing capacity
was transmission through cables,
carrying ten times as many circuits as
open wires but at the cost of high
attenuation. In October 1925 a cable
opened between New York and
Chicago, but with delicate and precise
construction pushing the limits of
the medium. Success came at great cost
in machinery and material, requiring
an expensive, low-resistance cable and
extensive loading and repeater
equipment.28 Making long cables practicable and economical required
numerous repeaters and vast numbers of
technicians distributed along the
route to maintain the delicate devices.
A simple comparison clarifies the
difficulties of both carrier and cable
transmission: the original (open-wire)
phone design. Also see Lloyd
Espenschied, “Application of Radio to Wire Transmission
Engineering,” Bell System Technical Journal 1 (October 1922) 117–41. On “wired
wireless,”
see Fagan (n. 9 above), 282.
26. E. H. Colpitts and O. B. Blackwell,
“Carrier Current Telephony and Telegraphy,”
Journal of the American Institute of
Electrical Engineers 40
(1921): 301–15, has a detailed
history of carrier methods in
telephony, as well as an elegant explanation of carrier modulation
and transmission.
27. In 1924 the type C system went into
service, incorporating lessons from the more
experimental A and B systems. Type C
carrier systems were so successful the last one was
not removed from service until 1980.
O’Neill (n. 2 above), 3–14.
28. See Charlesworth.
FIG. 1 Spectrum of a voice-band signal
modulated onto a carrier.
transcontinental line used fewer than
ten repeaters across the continent; a
carrier system over the same distance
needed forty, a cable would require
two hundred, and carrier and cables
combined would need even more.29
Hence carrier and cable transmission
required amplifiers of extremely high
quality.
An ideal amplifier is a pure
multiplier, taking an input signal and multiplying
it by some number (called gain) to produce an output. In other
words, a perfect amplifier has a linear
relationship between input and output.
On a graph of output versus input, the
amplifier’s response is literally
a straight line whose slope is the gain
(it might also have a frequency
dependent time delay, called phase shift, which is measured as an angle
between input and output sine waves).
For a vacuum tube, however, the
output versus input curve tends to be
more S-shaped (fig. 3). This nonlinearity
introduces distortion and causes two
problems. First, if the signal is
modulated on a carrier the nonlinearity
produces extraneous harmonics
outside of the desired signal band.
This becomes a problem with several signals
carrier-modulated onto the same wire.
The harmonics from one channel
overlap the bands of others, causing cross
talk—one conversation
bleeding through into another (fig. 4).
Second, since each amplifier adds a
little distortion, a long line with
numerous repeaters can garble the speech
beyond recognition. Thus, for BTL, as
the line became longer and longer,
and as more and more signals squeezed
onto a single wire, the amplifiers
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
416
29. O’Neill, 63, table.
FIG. 2 Carrier modulation.
MINDELLK|KRethinking
Feedback’s Myth of Origin
417
had to become correspondingly higher in
quality. This problem became a
high priority for Bell Labs at its
founding.
The Search for a Linear Amplifier
The first approach was to make the
vacuum tubes themselves more linear.
It was to this problem that Harold
Black turned his attention when he
joined the systems development
department of Western Electric in 1921. A
Massachusetts native, he had graduated
that year from Worcester
Polytechnic Institute in electrical
engineering. At Western Electric Black
worked with Mervin Kelley and the
vacuum-tube department, but with lit-
FIG. 3 Typical vacuum-tube nonlinearity. The
output anode
current is not a linear function of the
input grid voltage.
FIG. 4 Nonlinear amplifier causing distortion
and cross talk in a carrier system.
30. This account is based on Black,
“Inventing the Negative Feedback Amplifier” (n.
1 above), and Harold S. Black to A. C.
Dickieson, 16 June 1974, AT&T archives,Warren,
N.J. For a typical effort to design
linear vacuum-tube amplifiers, see E. W. Kellogg,
“Design of Non-Distorting Power
Amplifiers,” Electrical
Engineering 44
(1925): 490.
31. Harold S. Black, U.S. Patent No.
1,686,792, “Translating System.”
32. This assumption holds to within
1/µ, so if the amplifier gain is 100, then 1 percent
of the gain is determined by the vacuum
tube and 99 percent by the feedback network.
33. Black, “Inventing the Negative
Feedback Amplifier.”
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
418
tle success.Vacuum tubes, despite their
utility as circuit elements, remained
subtle, unruly—and nonlinear—devices
(hence Kelley’s efforts, years later,
overseeing the development of the
transistor).30
Black began to rethink the problem in
terms of signals. He conceptualized
the output of the amplifier as containing
a pure, wanted component,
the signal, and an impure, unwanted
component, the distortion. The problem,
then, was to somehow separate the two
and keep only the pure signal.
He came up with a clear, if inelegant,
solution: a “feed-forward” amplifier
that generated a copy of its own
distortion and subtracted it from the output
signal. Black built a laboratory
prototype that achieved the desired
result, and he applied for a patent in
1925.31 Though this setup proved that
a low-distortion amplifier was possible,
it was far from practicable. Black’s
overly complex new amplifier required
careful attention and continuous
adjustment, which engineers could do in
a testing lab but not for a system
deployed in the field.
STABILIZING BLACK’S BOX
For three years Black struggled to
simplify his solution. Finally, in 1927,
he had the epiphany on the ferry: if
the gain of the amplifier were reduced
by some amount, and that amount fed
back into the input, the linearity
could be greatly improved. In fact,
distortion was reduced (that is, linearity
improved) by the same factor by which
the gain was reduced. Black published
a simple explanation of the idea in a
1934 paper (fig. 5), showing that
the gain of the amplifier depends
primarily on the feedback network, , and
not on the gain, µ, of the amplifier
itself.32 The feedback network can consist
of only passive elements, such as
resistors, capacitors, and inductors,
that are both more linear than vacuum
tubes and more stable with respect
to temperature and other changes over
time. Consider an example: a feedback
amplifier with a vacuum-tube gain of
100,000 is enclosed in a feedback
loop that reduces its gain to 1,000.
The linearity of the amplifier overall
thus increases by a factor of 100, an
incredible improvement. The price,
of course, is to throw gain away and
settle for a much reduced level of
amplification. On 29 December 1927,
Black and BTL engineers succeeded
in making a feedback amplifier whose
distortion was reduced by a factor of
100,000 (and whose gain was reduced
accordingly).33
34. Ibid., 59–60.
35. Black to Dickieson; Harold S.
Black, patent application 298,155, 8 August 1928;
“File History of Black Application
Serial No. 298,155,” AT&T archives.
MINDELLK|KRethinking
Feedback’s Myth of Origin
419
Still, Black had no easy time
convincing others at Bell Labs of the utility
of his idea. He recalled that Jewett
supported him, but that the director
of research, Harold Arnold, refused to
accept a negative-feedback amplifier
and directed Black to design conventional
amplifiers instead.34 Black had
similar difficulties with the U. S.
Patent Office. His application for a “Wave
Translation System,” originally filed
in 1928, was not granted until 1937.35
To a generation of engineers who had
struggled to make the vacuum tube
amplify at all, throwing away the
hard-won gain seemed absurd.
More important, no one could understand
how an amplifier’s output
could be fed back to its input without
a progressive, divergent series of
oscillations. Bell engineers at the
time found it difficult to make a high-gain
amplifier without feedback. Subtle, uncontrolled feedback
paths would
arise through unintentional effects
such as stray capacitance between wires,
or even between elements within the
tube itself, and cause the amplifier to
go into “parasitic oscillation” or
“singing” (much like the whistling in a
FIG. 5 Harold Black’s negative feedback
amplifier. (Harold S. Black, “Stabilized
Feedback Amplifiers,” Bell System Technical Journal 13 [January 1934]: 3.)
poorly tuned public address system). In
1924, for example, two BTL engineers,
H. T. Friis and A. G. Jensen, studied
what they called “feed-back or
regeneration” as it occurred through a
tube, noting that it “makes the total
amplification vary irregularly in a
very undesirable manner and also makes
the set ‘sing’ at certain frequencies.”36 Black’s work went against the grain
for experienced amplifier designers:
they sought to eliminate feedback, not
to incorporate it.
Black interpreted the resistance to his
ideas as evidence of their radical
nature. Yet he was an engineer with a
bachelor’s degree in the systems
department; he did not possess the
analytical sophistication, the communications
skills, or the prestige of the research
scientists at BTL. His lab
assistant during this period, Alton C.
Dickieson, recalled Black as clashing
constantly with his own management and
the rest of BTL. Dickieson’s recollections
of Black’s troubles parallel the
inventor’s own accounts, so his
memory seems credible.37 Such conflicts were one thing for a lucid genius,
but Black was far from eloquent.“A
compulsive, non-stop talker,” Dickieson
recalled, Black “was inventive and
intuitive, but not particularly clear at
exposition.” His negative-feedback
circuit was only the latest in a series of
attempts over a period of several
years, all of which Dickieson wired up and
built, but, as he recalled, “none of
the schemes we tried showed any real
promise.” Dickieson also remembered
“quite a bit of rivalry” between the
Ph.D.-trained researchers and the
systems people. “There seemed to be
some feeling that exploratory development was the exclusive province
of the
research people. Mathematicians such as
Thornton Fry [head of BTL’s
math department] found Black’s
mathematics beneath contempt.”38
Black—restless, creative, and a bit
arrogant—was traversing the established
boundaries of the organization, and
running headlong into the cultural differences
between the research department and his
own lower-status systems
department.
Credible as Dickieson’s recollections
seem, no contemporary accounts
exist to support or refute them. The
documents do allow, however, a thorough
analysis of Black’s ideas, and how
Black himself had to transform
them (and enlist others to transform
them) in order to win their acceptance.
A key point involves his claim that the
epiphany on the ferry included
a concern for dynamic stability, that
if he “kept the device from oscillating
(singing, as we called it then)” it
would work. He implies that he understood
“stability” of the amplifier as the
central problem. But a look at
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
420
36. H. T. Friis and A. G. Jensen, “High
Frequency Amplifiers,” Bell
System Technical
Journal 3 (April 1924).
37. See, for example, Black, “Inventing
the Negative Feedback Amplifier” (n. 1
above) 59–60, for Black’s conflict with
H. D. Arnold and intimations of constant friction
with his superiors.
38. A. C. Dickieson to M. J. Kelley, 6
July 1972, AT&T archives, 43 09 03. Emphasis
added.
39. Friis and Jensen, 204.
40. Dickieson to Kelley.
41. “When many amplifiers are worked in
tandem . . . it becomes difficult to keep the
overall circuit efficiency constant,
variations in battery potentials and currents, small
when considered individually, adding up
to produce serious transmission changes in the
overall circuit”; Harold S. Black,
“Stabilized Feedback Amplifiers,” Bell System Technical
MINDELLK|KRethinking
Feedback’s Myth of Origin
421
Black’s conception of stability at the
time reveals it to be different from the
standard meaning of freedom from
oscillation. In fact, Black’s conceptions
of both negative feedback and stability
differed markedly from those of
much of the engineering community at
the time, although they would have
been familiar to engineers working on
the telephone network.
TWO CULTURES OF FEEDBACK AND STABILITY
Today the “negative” in
negative-feedback amplifiers means that the
feedback signal subtracts from the
input signal rather than adding to it
(that is, the sign of the feedback
signal is reversed). James Watt’s flyball governor
on a steam engine offers an analogy:
when the engine speeds up, the
spinning balls slow it down; when the
balls spin slower, they speed up the
engine. Hence the feedback is negative.
In Black’s time, however, the
definition of this specific-sounding term,
“negative feedback,” had yet to be
settled. The idea of positive feedback had
become current in the 1920s with the
introduction of the regenerative
amplifier. Positive feedback, or
regeneration, in a radio amplifier increased
the sensitivity of a receiving tube by
sending a wave back through an amplifier
many times. Black insisted that his
negative feedback referred to the
opposite of regeneration: gain was
reduced, not increased. Yet, to return to
the analogy of the steam engine
governor, Black’s use of “negative” means the
energy required to spin the balls
reduces the energy output of the engine, not
that the balls trigger an action that
slows it—hardly a significant effect for a
steam engine. In their 1924 paper Friis
and Jensen had made the same distinction
Black had between positive feedback and
negative feedback, that is,
distinguishing one from the other not
by the sign of the feedback itself but
rather by its effect on the amplifier’s
gain.39 In contrast, Nyquist and Bode,
when they built on Black’s work,
referred to negative feedback as that with
the sign reversed. Black had trouble
convincing others of the utility of his
invention in part because confusion
existed over basic matters of definition.
Misunderstanding also arose over the
critical idea of stability. Dickieson
recalled why those concerned with
singing in amplifiers did not take Black
seriously: “Harold did not even
approach the question of stability—he simply
assumed that it did not sing.”40 Actually, Black was deeply concerned
with stability: his first published
paper on the amplifier appeared in 1934
with the title, “Stabilized Feedback
Amplifiers.” But to Black “stability”
referred not to freedom from oscillation
but to the long-term behavior of
components in the telephone network.41 Life in the network exposed a tele-
phone repeater to a harsh world, and
Black sought to insulate the signal
from brutal reality.He wanted to use
feedback to stabilize the characteristics
of the amplifier over time. Temperature
changes, aging of components,
changes in the power supply, and any
number of other factors could affect
the performance of an amplifier. Rain
and temperature fluctuations, for
instance, changed the resistance of the
wire and caused significant variations
in attenuation, sometimes by a factor
of a hundred or more over the course
of a single day, and to comparable
degree across the change of seasons.42
These fluctuations could greatly alter
the physics of transmission, a potentially
disastrous effect for systems operating
close to their physical limits.
Yet to the scientifically trained
engineers at BTL, stability, in the sense
of freedom from oscillation, was the
main difficulty of the feedback amplifier.
Homer Dudley, discussing Black’s paper
in the journal Electrical
Engineering, listed freedom from singing as one of
the two most important
problems for the amplifier.43 Yet this type of stability was not Black’s concern.
His original patent application, filed
in 1928, makes no mention of
even the possibility of singing or
oscillation.44 When resubmitting the
application in 1932, he added this
clarification: “Another difficulty in
amplifier operation is instability, not
used here as meaning the singing tendency,
but rather signifying constancy of
operation as an amplifier with
changes in battery voltages,
temperature, apparatus changes including
changes in tubes, aging, and kindred
causes . . . Applicant has discovered
that the stability of operation of an
amplifier can be greatly improved by
the use of negative feedback.”45 Black even acknowledges the other meaning
of stability, but assigns it
unequivocal second billing: “Applicant uses
negative feedback for a purpose quite different
from that of the prior
art
which was to prevent self-oscillation
or ‘singing.’To make this clearer, applicant’s
invention is not concerned, except in a very secondary way . . . with the
singing tendency of a circuit. Its
primary response has no relation to the
phenomena of self-oscillation”
(emphasis added).46 In the patent, Black
“simply assumed” that the amplifier did
not oscillate.
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
422
Journal 13 (January 1934). This paper was
presented at the winter convention of the
American Institute of Electrical
Engineers,New York, January 1934, and also published in
Electrical Engineering 53 (January 1934): 114–20. See also the
discussions of the paper in
Electrical Engineering by F. A. Cowan (April 1934): 590; G.
Ireland and H. W. Dudley
(March 1934): 461–62; and H. Nyquist
(September 1934): 1311–12.
42. H. A. Affel, C. S. Demarest, and
C.W. Green, “Carrier Systems on Long Distance
Telephone Lines,” Bell System Technical Journal 7 (July 1928): 384. Green was Harold
Black’s boss.
43. Dudley, discussion of Black,
“Stabilized Feedback Amplifiers.”
44. Harold S. Black, patent application
298,155; “File History of Black Application
Serial No. 298, 155” AT&T archives.
45. Harold S. Black, U.S. Patent No. 2,102,671,
“Wave Translation System,” 2.
46. Ibid.
47. In 1921, for example, Colpitts and
Blackwell wrote that singing in a carrier system
could arise when the gain was greater
than one and when there existed “sufficient
unbalance” between the circuits. Colpitts
and Blackwell (n. 26 above), 313.
48. In 1926 Harvey Fletcher analyzed
the howling telephone as a dynamic electrical
system to understand the relationship
between impedance, frequency, and the tendency
to break into the oscillation; “The
Theory of the Operation of the Howling Telephone
with Experimental Confirmation,”Bell System Technical Journal 5 (January 1926): 27–49.
Fletcher’s paper does not employ the
terms “stability” or “feedback” in its analysis,
although it does analyze
electro-acoustic circuits that greatly resemble canonical feedback
systems. Shaw (n. 6 above), 382–83. On
the problems of handset howling, see Fagan
(n. 9 above), 146–50.
49. Bennett (n. 2 above), 77. See also
Ronald M. Foster, “A Reactance Theorem,” Bell
System Technical Journal 3 (April 1924): 266.
MINDELLK|KRethinking
Feedback’s Myth of Origin
423
Black’s conception of stability,
strange as it may seem, derived from his
position in the systems development
department as opposed to the
research department. Where a researcher
might focus on the theoretical
behavior of the system, Black was
concerned with its concrete, daily characteristics.
To system engineers such as Black,
“stable” amplifiers were those
that retained consistent performance in
the face of the varying conditions
experienced by equipment in the
telephone network. Consistency, regularity,
and stability of the circuit elements
were critical to transmission systems.
Black employed this operational
conception of stability in the analysis
of his amplifier. He used the term
stability as an engineer who saw the
system as a concrete, operational
entity, not as one who thought in abstract
diagrams.
Nevertheless, system engineers, despite
their emphasis on transmission
stability, should also have been
familiar with dynamic stability. Repeater
amplifiers had always had problems with
singing; they would sing if the signal
from one direction of transmission
leaked into the other (a full repeater
requires two amplifiers, one for each
direction of transmission). In
response to these problems, telephone
engineers filtered out the singing frequencies
and limited the amount of gain in each
repeater. Carrier systems
also tended to sing, either locally or
through the transmission line.47 The
now familiar telephone handset,
introduced in the late 1920s, depended on
understanding and preventing the
singing or “howling” that resulted from
the mouthpiece picking up sound from
the earpiece.48 Moreover, the stability
of motion had been a popular topic in
physics in the late nineteenth
century, and at least some telephone
engineers in the 1920s were aware of
it, although they were unsure of its
relevance to vacuum-tube circuits.49
Multiple, overlapping conceptions of
negative feedback and stability
thus surrounded the introduction of Black’s
amplifier. The Bell Laboratories
research culture was not monolithic,
but rather comprised at least
two engineering subcultures:
Ph.D.-level mathematicians and scientists
interested in fundamental questions,
and system engineers such as Black,
concerned with building the network and
keeping it running. Their differing
backgrounds, and differing notions of
ideas such as “stability,” help
explain why the research department did
not take Black seriously. As
Nyquist and Bode’s contributions make
clear, it would take both approaches
to make the feedback amplifier a
practical reality.
When Black invented the
negative-feedback amplifier, he invented a
different machine from both the one it
eventually became and the one he
remembered. Especially in light of his
claim that he recognized feedback as
a unifying principle across different
types of systems, these clashing visions
raise the question of whether Black
drew on the long tradition of regulators
and governors that preceded him.
SINGING AND HUNTING
Feedback techniques had of course been
commonly used for a long
time in governors, regulators,
thermostats, automatic pilots, and numerous
other devices. In his memoirs, Black
said that he understood his feedback
amplifier as part of that technological
trajectory. The significance of the
origin myth rests on Black’s supposed
recognition that negative feedback is
isomorphic across diverse types of
systems. Indeed, Black’s patent, as
issued, states that the
negative-feedback principle applies to more than
electronic amplifiers: “the invention
is applicable to any kind of wave transmission
such as electrical, mechanical, or
acoustical . . . the terms used have
been generic systems.” But the patent
never specifies what those other
applications might be, and a steam-engine
governor, an automatic pilot, or
a servomechanism fit only loosely into
the category “wave translation system”
(the title of Black’s patent). Black
likely had in mind more directly
analogous systems, such as the numerous
electro-acoustic translations
required in telephony. Neither the
patent, nor any of Black’s early writings,
nor the writings of any of the BTL
feedback theorists for at least ten years,
mention regulators, governors,
automatic pilots, or any of the myriad
devices we now understand as employing
negative feedback.
Nonetheless, such devices were
themselves in wide use within the telephone
network. Telephone repeaters needed
regular adjustment as the
characteristics of the transmission
lines changed in response to environmental
changes. In the late 1920s AT&T
installed automatic regulators in
about every fourth repeater station;
these devices adjusted amplifier gain in
response to a feedback loop that sensed
the wire’s characteristics. In 1929,
for example, the New York/Chicago line
included six regulating stations
among its twenty repeaters.50 In this light, Black’s stability of transmission
was a kind of automation: his stable
feedback amplifiers relieved network
maintenance personnel of the task of
adjusting the delicate amplifiers.51
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
424
50. E. D. Johnson, “Transmission
Regulating System for Toll Cables,” Bell Laboratories
Record 7 (January 1929): 183–87.
51. Ireland (n. 41 above).
52. Hugh M. Stoller, “Synchronization
and Speed Control of Synchronized Sound
Pictures,” Bell System Technical Journal 8 (January 1929): 184–95. Also see H.
M. Stoller
and E. R. Morton, “Synchronization of
Television,” Bell
System Technical Journal 6
(October 1927): 604–15, and H. M.
Stoller “Speed Control for the Sound-Picture
System,” Bell Laboratories Record 7 (November 1928): 101–5. W. Trinks, Governors and
the Governing of Prime Movers (New York, 1919).
53. H. M. Stoller, “Speed Control for
the Sound-Picture System,” Bell Laboratories
Record 7 (November 1928): 101–5. Stoller also
published on voltage regulators; H. M.
Stoller and J. R. Power, “A Precision
Regulator for Alternating Voltage,”Transactions of the
American Institute of Electrical
Engineers 48
(1929): 808–11.
MINDELLK|KRethinking
Feedback’s Myth of Origin
425
Regulators and governors could also be
found within BTL’s engineering
culture. Sound movies, for example,
required tight control lest variations in
film speed change the pitch of the
sound and become annoying to the
viewer. Similarly, early television
systems in development at BTL in the
1920s employed large mechanical disks
to scan the picture (instead of the
later electron beams). Keeping these
disks exactly aligned required precise
regulators. In a series of papers
published between 1927 and 1929, Hugh
Stoller of the apparatus department
explicitly compared his speed controls
to steam engine governors and even
discussed the phenomenon of “hunting,”
equivalent to singing in an amplifier.52 He included a drawing of a flyball
governor in the Bell Laboratories Record, and used “stability” in the
sense of freedom from oscillation.
Stoller even used the term “feed back”
for the electrical speed regulation in
his own circuits.53 Had Black looked,
he would have found discussion of
traditional mechanical regulators in his
own organization and its publications
In fact, the analogy between a
mechanical regulator and an electronic
one would not have been a great leap
for Black, as Stoller made the connection
clearly but without much fanfare. But
Black did not take that step.
He did not see his negative-feedback
amplifier as analogous to regulators
and governors and he did not see
hunting in those devices as comparable
to singing in an amplifier.
This critical look at Black’s
conception of his amplifier provides some
perspective on the origin myth. Black’s
flash of insight, however much it
enlightened him on the structure of
negative feedback, did not give him an
artifact he could sell, nor did it give
him the modern conception of a negative-
feedback amplifier or a broadly
applicable notion of feedback. But it
would be wrong to suggest that Black
would have found a more receptive
audience for his invention had he realized
that the amplifier’s stability was
a key problem, that negative feedback
worked similarly to regulation, that
singing resembled hunting. These
judgments we can only make with hindsight.
The important historical point must be
made positively: to Black the
amplifier was a means of throwing away
gain to achieve linearity in a vacuum
tube, a way of stabilizing the
repeaters in the telephone system subject
to variation and hazard. On these
points he was always clear, consistent,
and determined.
In his 1934 paper “Stabilized Feedback
Amplifiers,” Black presented his
amplifier to the world. He attributed
the delay from his 1927 insight to the
1934 paper to corporate secrecy, but
that can account for at most five of the
seven years. Black’s paper, in fact, was
not the first word from the telephone
company on the negative-feedback
amplifier; that word, a paper that Black
cited and discussed, had appeared two
years earlier. It was the work of an
ally, to whom Black had turned for
help, but who remade Black’s box.Harry
Nyquist rethought negative feedback by
redefining stability.
DIAGRAMMING STABILITY
Harry Nyquist, a Swedish immigrant with
a Ph.D. in physics from Yale
University, brought negative feedback
from Black’s curiosity into the network.
Nyquist belonged not to BTL but to the
development and research
department of AT&T; he stabilized
Black’s box by bringing it into the frequency
domain.54
In May 1928 Nyquist asked Black to join
in developing a new carrier
system and to include the
negative-feedback amplifier in a demonstration
of new transmission techniques. This
project, known as the Morristown
Trial, installed seventy-eight
repeaters of Black’s design spaced every
twenty-five miles of cable. The cable
folded back on itself, so all the amplifiers
were located in the same laboratory in
Morristown, New Jersey.55
Before his work on the Morristown
trial, Nyquist had worked on problems
of both transmission stability and
regulation.56 With the Morristown trial,
Nyquist brought this experience to
amplifiers.
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
426
54. Hendrik W. Bode, “Harry Nyquist”
(obit.), IEEE
Spectrum 14
(April 1977).
55. For a detailed account of the
Morristown Trial, see A. B. Clark and B.W. Kenall,
“Carrier in Cable,” Bell System Technical Journal 12 (July 1933): 251–62; see also
O’Neill
(n. 2 above), chap. 5, “Carrier on
Cable.”Making the system work as planned proved no
simple matter, but such was the purpose
of an engineering trial. Repeater amplifiers did
not pose the only problems: cable
design (the number, size, and shielding of each of the
many wire pairs) proved especially
critical as well. Shielding, grounding, and interference
between signals plagued the system.
Because of the depression, AT&T changed its
emphasis from new systems to improving
capacity with the existing plant. Engineers at
BTL had several years to refine the
results of Morristown and to work on ways of compressing
more transmission onto existing wires.
The Morristown Trial formed the basis
for the K-type carrier system,
introduced in the late 1930s, which carried twelve voice
channels on cables at frequencies from
12 to 50 kHz for distances up to 4,000 miles. Kcarrier
furnished 70 percent of the increased
capacity in the country (which doubled
from 1940 to 1947) and remained in
service until at least 1980. K-carrier also included a
pilot wire-transmission regulation
scheme, with an automatic self-balancing regulator
and a self-synchronizing motor. C.W.
Green and E. I. Green, “A Carrier Telephone System
for Toll Cables,” Bell System Technical Journal 17 (January 1938).
56. H.Nyquist,U.S. Patent No.
1,887,599, “Constant Current Regulation”; U.S. Patent
No. 1,683,725, “Phase Regulating
System.” Applications filed in 1928 and 1926, respectively.
B. P. Hamilton, H. Nyquist, M. B. Long,
W. A. Phelps, “Voice-Frequency Carrier
Telegraph System for Cables,” Transactions of the American Institute
of Electrical Engineers
44 (February 1925): 327–39. This paper
(which erroneously gives Nyquist’s first initial as
N.) also includes a discussion of the
precision governor required for generating carrier
frequencies for this telegraph system,
suggesting that Nyquist had exposure to regulation
before his 1932 paper on feedback,
“Regeneration Theory,” Bell
System Technical Journal
11 (January 1932): 126–47.
57. Nyquist, discussion of Black,
“Stabilized Feedback Amplifiers” (n. 41 above).
58. Nyquist, “Regeneration Theory”;
Bennett (n. 2 above), 82–84.
MINDELLK|KRethinking
Feedback’s Myth of Origin
427
His 1932 paper, “Regeneration Theory,”
provided a rigorous set of
measurable conditions by which to
determine the stability of a feedback
amplifier. He redefined feedback as a
frequency-dependent phenomenon,
and stability in terms of transient
disturbances composed of different frequencies
(essentially shocks to the system).
“For the purpose of studying
the singing condition,” he wrote, “it
is permissible to regard the feed-back
phenomenon as a series of waves.”57 For Nyquist, if all disturbances die out
after a finite period of time, the
circuit is stable. If any disturbance goes on
indefinitely, the circuit is unstable
(fig. 6).58 In light of this definition, it
became clear to Nyquist that two
conditions are necessary and sufficient to
make an amplifier unstable and cause
singing: first, if the wave coming
around the feedback loop equals or
exceeds in magnitude the input to the
amplifier, that is, if the gain is
equal to or greater than one; second, if the
FIG. 6 “For the purpose of studying the
singing condition, it is permissible to
study the feedback condition as a
series of waves. . . .” (H. Nyquist, discussion
of a paper by H. S. Black, “Stabilized
Feedback Amplifiers,” Electrical
Engineering
53 [September 1934], 1311.)
feedback wave is inverted compared to
the input wave (that is, its phase
shift is 180°). If these conditions are
both met for any frequency, then the
amplifier is unstable and will
oscillate.Nyquist turned these conditions into
a simple, empirical method for
determining stability: open the loop, measure
the amplifier’s parameters (gain and
phase shift) for varying frequencies,
record them on a polar plot, and use
the plot to graphically determine
stability (fig. 7).59 This plot became known as a “Nyquist diagram,” and
the
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
428
59.Nyquist’s method was this: First,
break the loop so the amplifier will not feed back
on itself. Then measure its “open loop
characteristics,” plotting two easily measured quantities,
gain and phase, against each other as
they vary with the frequency of the input signal.
If the resulting curve encloses the
point that represents a unity gain and 180° shift, the
system is unstable. If the point lies
outside the curve, the system is stable. Nyquist,
“Regeneration Theory.” In 1934, BTL
engineers compared Nyquist’s criterion to Routh’s
test from his 1877 Adams Prize paper on
stability in dynamic mechanical systems. They
FIG. 7 Original-style Nyquist diagram, showing
gain (magnitude) versus phase
shift (angle) plotted for several
different frequencies on a polar plot. Since the
curve does not enclose the point (1,0),
the system is stable. If curve did enclose
that point, the system would be
unstable. (After H. Bode, “Feedback: The History
of an Idea,” reprinted in Selected Papers on Mathematical Trends
in Control
Theory, ed. Richard Bellman [New York, 1964],
114.)
test remains the “Nyquist stability
criterion” or the “Nyquist criterion.”60
This technique reduced a significant
amount of calculation to a simple procedure,
a literary technology, and a tool for
engineers to think with. It is still
used today.
FEEDBACK AS A NETWORK PROBLEM
It remained for one more BTL engineer,
Hendrik W. Bode, to complete
telephony’s prewar phase of feedback
theory. Bode came to BTL in 1926,
fresh from a master’s degree at Ohio
State, where had also done his undergraduate
degree; he received a Ph.D. in physics
from Columbia in 1935.
Bode’s expertise was not in feedback,
nor even in amplifiers or vacuum
tubes, but in the useful but esoteric network theory. The theory of electrical
networks dealt not with the telephone
network itself but with abstractions
of the numerous small networks of
resistance, capacitance, and inductance
that determined its behavior.61
As the Bell System adopted carrier
transmission and began to manipulate
signals in the frequency domain,
electrical networks became increasingly
critical. Filter networks, for example,
separated specific frequencies
out of the spectrum, and equalizer
networks compensated for the distortion
in a transmission line. In 1934, Bode
developed and published a general
theory that accounted for all types of
networks. Bode called this work
“a sort of algebra” that allowed designers
to manipulate network designs
graphically, without solving their
tangled equations.62
Bode’s work on networks merged with
feedback amplifiers because of
yet another new transmission medium,
coaxial cable, which had only one
conductor surrounded by a conductive
shield. These cables allowed several
hundred conversations to be multiplexed
together and could also carry the
new broadband television signals. As
with the jump from open wire to
cable, the jump to coaxial cables
placed heavier demands on repeaters,
equalizers, and system performance
overall.63
MINDELLK|KRethinking
Feedback’s Myth of Origin
429
found the two stability analyses
compatible, and thus linked the new feedback theory to
the older work on dynamic stability. Despite
this link, however, their work makes no mention
of applying feedback amplifier theory
to other dynamic systems. E. Peterson, J. G.
Kreer, and L. A. Ware, “Regeneration
Theory and Experiment,” Bell System Technical
Journal 13 (October 1934): 680–700.
60. Bennett, 83.
61. S.Millman, ed., A History of Engineering Science in the
Bell System: Communications
Sciences (1925–1980) (Murray Hill, N.J., 1984), 16–17. Also
see O’Neill (n. 2 above),
204–8. For a good summary of the work
on network theory in the twenties and thirties,
see Karl L. Wildes and Nilo A.
Lindgren, A
Century of Electrical Engineering and Computer
Science at MIT, 1882–1982 (Cambridge, 1985), chap. 9, “Network
Analysis and
Synthesis: Ernst A. Guillemin.”
62. H. W. Bode, “General Theory of Electric
Wave Filters,” Journal
of Mathematics
and Physics 13 (November 1934): 275–362.
63. L. Espenschied and M. E. Strieby,
“Systems for Wide-Band Transmission over
In 1934, Bode set about designing an
equalizing network for a repeater
amplifier for coaxial cable.64 The trouble was, Bode had to design the equalizer
network after the amplifier had already
been designed, and such post
hoc modification made the amplifier
unstable. “I sweated over this problem
for a long time without success,” Bode
recalled. Finally, “in desperation,” he
redesigned the entire amplifier using
techniques from network theory.
Where Nyquist had provided a way to
determine if an existing amplifier
was stable, Bode now aimed to design a
stable amplifier to meet specified
parameters for performance.
Bode’s 1940 paper “Relations Between
Attenuation and Phase in Feedback
Amplifier Design,” remains his
best-known and most succinct contribution
to feedback theory. The opening pages
have a decidedly pessimistic
tone, as Bode notes that the stability
of a feedback amplifier “is always just
around the corner.”He begins: “The
engineer who embarks upon the design
of a feedback amplifier must be a
creature of mixed emotions. On the one
hand, he can rejoice in the improvements
in the characteristics of the structure
which feedback promises to secure him.
On the other hand, he knows
that unless he can finally adjust the
phase and attenuation characteristics
around the feedback loop so the
amplifier will not spontaneously burst into
uncontrollable singing, none of these
advantages can be actually realized.”65
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
430
Coaxial Lines,” Bell System Technical Journal 13 (October 1934): 654–79. M. E.
Strieby, “A
Million-Cycle Telephone System,” Bell System Technical Journal 16 (January 1937): 1–9.
See also O’Neill, chap. 6, “Coaxial
Cable,” especially 131–39. The system Bode worked on
became known as the L1; it was tested
on a line from New York to Philadelphia in
1936–38 and put into service just
before the war.
64. Here Bode came to a critical
realization. The overall amplifier behaves like the
reciprocal of its feedback elements—when the
feedback path divides, for example, the
amplifier overall multiplies, when the
feedback element passes certain frequencies, the
amplifier overall blocks those
frequencies, and vice versa. A passive equalizer had to
mimic the reciprocal of the
transmission line to cancel out its effects. In network theory,
however, creating the inverse of a physical
network could be a complicated affair, and
might not even be physically possible.
Bode realized, however, that since the feedback
amplifier inverted the behavior of the
feedback network, the problem of equalizer design
reduced to the simpler problem of
designing a feedback network to simulate the transmission
line exactly, rather than to invert it.
H. W. Bode, “Variable Equalizers,” Bell
System Technical Journal 17 (April 1938): 229–44. Black wrote in
1934: “For many types
of frequency characteristics it is
difficult, and for some impossible, to construct a passive
network having the exact inverse
characteristic [as the transmission line].With this type
of [feedback] amplifier, however, it is
only necessary to place in the feedback circuit
apparatus possessing the same
characteristic as that to be corrected.” Black, “Stabilized
Feedback Amplifiers” (n. 41 above),
294.
65. Bode, “Relations Between
Attenuation and Phase in Feedback Amplifier Design”
(n. 1 above). For other discussions of
this paper, see Bennett (n. 2 above), 84–86;Millman,
29–30; O’Neill, 68–70. In later years,
Bode displayed some aversion to Black’s version of
events. He wrote to A. C. Dickieson in
1974, after reviewing Black’s account, that “this is
not exactly how one ordinarily writes
formal technical history [interestingly, Bode had
some notion of ‘formal technical
history’]. . . . Have you thought of a less personalized
treatment in which pieces of Black’s
account are woven in with expository text of your
own? . . . It might be possible to
eliminate, for example, the references to Steinmetz and
Hartley, which seem to me to be
irrelevancies. In a less personalized account, it might be
possible to present basic technological
issues in a more satisfactory way. For example, as
the paper now stands it seems to imply
that Black deserves credit for the pioneer investigation
of nonlinear effects in long systems. I
doubt whether this is really accurate. . . . I was
also a little disturbed by Harold’s
claim that he outfaced the U. S. Patent office on every
one of 126 claims. I didn’t know that
the Patent Office gave ground that easily. In any case,
credit should probably go to the
long-suffering patent attorney who wrote all those letters.”
Bode to Dickieson, 17 September 1974,
AT&T archives.
66. Elsewhere he likened the feedback
amplifier designer to “a man who is trying to
sleep under a blanket too short for
him. Every time he pulls it up around his chin his feet
get cold.” H. W. Bode, “Design Method
for Feedback Amplifiers—Case 19878,” 1 May
1936, AT&T archives.
67. Bode, “Relations Between
Attenuation and Phase in Feedback Amplifier Design.”
68. Ibid., 426–35.
69. Bode, “Feedback: The History of an
Idea,” in Bellman (n. 2 above), 117.
70. H.W. Bode,Network Analysis and Feedback Amplifier
Design (New
York, 1945), iii.
MINDELLK|KRethinking
Feedback’s Myth of Origin
431
Bode likens a feedback amplifier to a
perpetual motion machine that
would work “except for one little
factor” that never quite goes away, despite
all the tweaking.66 Bode elucidates the parameters (gain and phase
shift)
“which impose limits to what can and
cannot be done in a feedback design
. . . and [forbid] the building of a
perpetual motion machine.” The price of
using feedback, he continues, “turns
out to be surprisingly high.” It “places
a burden on the designer,” and without
new tools “he is helpless.” Bode
seems to be addressing Black himself
and his uncritical exuberance for the
benefits of feedback, regardless of
stability problems. “Unfortunately, the
situation appears to be an inevitable
one. The mathematical laws are inexorable.”
67 Like Nyquist, Bode developed simple, graphical
techniques to
determine stability by plotting
observed and analytic quantities. Like
Nyquist diagrams, these graphs survive
today as “Bode plots.”
Nyquist’s stability conditions produced
an answer: the amplifier is stable
or it is not. Bode’s technique assessed
how much stability it had, with
quantitative measures. Bode also
imposed limits on the possible performance
of the feedback amplifier by proving
that “we cannot obtain unconditionally
stable amplifiers with as much feedback
as we please” because too
much feedback could make the amplifier
unstable.68 His name is permanently
associated with feedback, but he always
linked it to its network roots:
“it is still the technique of an
equalizer designer,” he wrote in retrospect. “I
can imagine that the situation may well
seem baffling to someone without
such a background.”69 The title of Bode’s 1945 book Network Theory and
Feedback Amplifier Design reflects his primary experience in
networks, with
secondary application to amplifiers.
During World War II, Bode and BTL
widely distributed the unpublished
manuscript to other laboratories working
on control systems.70 Bode acknowledged a certain amount of “unnec-
71. Ibid., iv.
72. H. Nyquist, “Certain Factors
Affecting Telegraph Speed,” Bell System Technical
Journal 3 (April 1924): 324–46. For telegraph
sampling, the main paper was H. Nyquist,
“Certain Topics in Telegraph
Transmission Theory,” Transactions
of the American Institute
of Electrical Engineers 47 (February 1928): 617–44. See also
the discussion of this paper by
Nyquist’s son-in-law, John C. Lozier,
“The Oldenberger Award Response: An Appreciation
of Harry Nyquist,” Journal of Dynamic Systems,Measurement
and Control 98
(June 1976):
127–28. Nyquist’s measure, that a wave
must be sampled at twice its bandwidth to be
transmitted without distortion, is
frequently referred to as “the Nyquist rate.” A modern
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
432
essary refinement” of the design
methods in the book, but explains that
they were required for telephone
repeater amplifiers, with their unusually
high standards for performance.71 Even today, through Bode’s plots, feedback
techniques retain the traces of the
network theory of the 1920s.
Speaking Machinery and the Transmission
of Information
The work done by Black, Bode, and
Nyquist brought negative feedback
and the vacuum tube within the realm of
signals, frequencies, and networks.
The high-quality linear repeater
amplifiers these men developed
furthered the separation of the message
inherent in the telephone signal
from the energy required to transmit it
down the line. Black’s feedback
amplifier aimed to regulate
transmission and insulate the performance of
the technical network from its physical
and meteorological environment.
Nyquist and Bode addressed the immediate
problems of frequency
response and dynamic stability. Because
self-regulation could rapidly turn
to oscillation, avoiding instability
became a primary concern of feedbackamplifier
design. Developing the feedback
amplifier connected at every
point to problems of the telephone
network, including long-distance transmission,
carrier modulation, and the role of
fundamental research in the
system overall.
Negative-feedback amplifiers evolved
together with a conception of the
network as a social device, and of machines
as active speech producers—a
vision actively supported by the new
research organization. Technically, this
vision incorporated both telegraphy and
telephony, text and speech (and
later images), into theories of
processing signals, manipulating them in the
frequency domain, and precisely
matching them to transmission channels.
Indeed, at the same time that Nyquist
was theorizing negative feedback he
was working out the relations between
bandwidth and channel capacity, the
interchangeability of telephone and
telegraph signals, and the effect of noise
on transmission rates. Nyquist also
attacked the problem of chopping up a
signal into discrete bits or “signal
elements,” transmitting them individually,
and then using them to reconstruct the
original signal. Today, Nyquist’s
“sampling theorem” still determines the
rates at which our analog world is
sampled and converted into digital
form.72 Similarly, BTL researcher Ralph
Hartley’s work proposed quantitative
measures for the transmission of signals
independent of their nature or content.
Nyquist and Hartley laid the
groundwork for the theory of
information that Claude Shannon would
articulate in 1948.73 And Homer Dudley, in an article titled “The Carrier
Nature of Speech,” explicitly compared
human language to network traffic.
At the Century of Progress Exposition
in Chicago in 1933, the AT&T exhibit
featured Dudley’s speech synthesizer
and promoted the company’s new
Teletypewriter services.74 Other BTL researchers during these years developed
automatic switches, talking movies,
stereophonic sound, artificial
organs for listening and speaking, and
“invisible orchestras” for transmitting
high-fidelity audio over the network.75 Each furthered, in its own way,
the abstraction of signals and the
extension of human activity by the telephone’s
spreading network.
Developments in communications theory
did not simply reflect technical
systems and facilitate their
convergence. They also supported AT&T’s
corporate goals. Frank Jewett, speaking
to the National Academy of Sciences
in 1935, rejected the distinctions
between types of signals: “We are
prone to think and, what is worse, to
act in terms of telegraphy, telephony,
radio broadcasting, telephotography, or
television, as though they were
MINDELLK|KRethinking
Feedback’s Myth of Origin
433
CD player, for example, samples music
at 44 kHz in order to reproduce it in the audible
band of about 20 kHz. For Nyquist’s
work on noise, see “Thermal Agitation of Electric
Charge in Conductors,” 110–13.
73. R. V. L. Hartley, “Transmission of
Information,” Bell
System Technical Journal 7
(July 1928): 535–63. See brief
discussions of Nyquist and Hartley by E. Colin Cherry, “A
History of the Theory of Information,” Proceedings of the Institution of Electrical
Engineers
98 (September 1951): 386, and by J. R.
Pierce, “The Early Days of Information Theory,”
IEEE Transactions on Information Theory, no. 1 (January 1973): 3. In his
foundational paper
on information theory, Shannon cited
Nyquist’s two papers on telegraph transmission,
“Certain Factors Affecting Telegraph
Speed” and “Certain Topics in Telegraph Transmission
Theory,” and Hartley’s “Transmission of
Information” in the first paragraph.
Claude Shannon, “A Mathematical Theory
of Communication,” parts 1 and 2, Bell System
Technical Journal 27 (July/October, 1948), 379–423,
623–56, reprinted in Claude
Elwood
Shannon: Collected Papers, ed. N. J. A. Sloane and Aaron D.Wyner
(New York, 1993), 5–83.
74. Homer Dudley, “The Carrier Nature
of Speech,” Bell
System Technical Journal 19
(October 1940): 495–515. “The Bell
System Exhibit at the Century of Progress Exposition”
Bell Laboratories Record 11 (July 1933).
75. Kenneth Lipartito, “When Women Were
Switches: Technology, Work, and
Gender in the Telephone Industry,
1890–1920,” American
Historical Review 99
(October
1994): 1074–111. Sheldon Hochheiser,
“What Makes the Picture Talk: AT&T and the
Development of Sound Motion Picture
Technology,” IEEE
Transactions on Education 35,
no. 4 (November 1992): 278–85. Harvey
Fletcher, “The Nature of Speech and Its
Interpretation,” Bell System Technical Journal 1 (July 1922): 129; “Physical
Measurements
of Audition and Their Bearing on the
Theory of Hearing,” Bell
System Technical Journal
2 (October 1923): 145; “Useful Numerical
Constants of Speech and Hearing,” Bell
System Technical Journal 4 (July 1925):
375–86. Robert E. McGinn, “Stokowski and the
Bell Telephone Laboratories:
Collaboration in the Development of High-Fidelity Sound
Reproduction,” Technology and Culture 24 (1983): 43.
things apart.” Jewett argued instead
that these technologies merely represented
different embodiments of a common idea
of communication.
“[T]hey are merely variant parts of a
common applied science. One and all,
they depend for the functioning and
utility on the transmission to a distance
of some form of electrical energy whose
proper manipulation makes
possible substantially instantaneous
transfer of intelligence.”76 Government
regulation persisted in making
distinctions between media (radio, telephony,
and so on), each controlled by their
own vested interests.When policy
followed science and treated all
signals as equivalent, Jewett argued, then
AT&T, with its natural monopoly,
would emerge as the unified communications
company: a builder of transmission, a
carrier of long-distance signals,
and a switcher of information.
Jewett’s vision echoed Theodore Vail’s
“one policy, one system” motto,
updated by the advances in technology,
theory, and the organization of
research at AT&T in the 1920s and
1930s. Feedback theory at Bell Labs contributed
to the rapidly converging ideas about
signals and communications
that Jewett articulated. It was in this
environment that Harold Black had his
vision of feedback on the Lackawanna
ferry in 1927.
Still, despite Jewett’s call to unify
communications, Black, Nyquist, and
Bode kept their ideas within the
existing network. They did not see their
contributions to feedback theory as
significant to the world of governors,
regulators, servomechanisms, or
automatic controls. Contrary to Black’s
recollection, the realization that
feedback described common phenomena
in a variety of settings did not
crystallize until World War II, when new
institutions brought engineers from
diverse backgrounds together to construct
military control systems. Only then
were the techniques developed at
BTL to deal with feedback, frequencies,
and noise applied to mechanical
and hydraulic systems, and to the human
operators themselves. Only then
did feedback become prominent as a
general principle in engineering, and
only afterward, with the work of
Norbert Wiener, Claude Shannon, and
numerous others, did Black’s, Bode’s,
and Nyquist’s ideas move beyond
amplifiers and into a broad range of
disciplines. Feedback is indeed fundamental
to our technological world, but Harold
Black’s epiphany, more than
a foundational moment, was one of a
series of technical insights that
allowed engineers to separate human
communications from their electrical
substrates, to send them through
geographically extensive networks, and to
represent the world in a common
language of signals.
T E C H N O L O G Y A N D C U L T U R E
JULY
2000
VOL. 41
434
76. Frank B. Jewett, “Electrical
Communication, Past, Present, and Future,” speech to
the National Academy of Sciences, April
1935, reprinted in Bell
Telephone Quarterly 14
(July 1935): 167–99.
In the Dec. 77 issue of IEEE Spectrum [1] Harold
S. Black wrote:
http://digilander.libero.it/paeng/feedforward_concepts.htm
Inventing the ‘black box’:
mathematics as a neglected enabling technology
in the history of
communications engineering
Chris Bissell
The Open University, UK
Milton Keynes MK7 6AA
http://www.ieee.org/organizations/history_center/cht_papers/bissell.pdf