Download “Beautiful and cantankerous instruments”: telescopes, technology

Exp Astron (2009) 25:79–89
DOI 10.1007/s10686-009-9138-9
REVIEW ARTICLE
“Beautiful and cantankerous instruments”: telescopes,
technology, and astronomy’s changing practice
W. Patrick McCray
Received: 5 January 2009 / Accepted: 6 January 2009 / Published online: 3 February 2009
© Springer Science + Business Media B.V. 2009
Abstract Between the dedication of the 200” Hale Telescope in 1948 and the
completion of today’s 8–10 m behemoths, astronomers’ most iconic symbol,
the telescope itself—its design, its technology, and its use—was transformed as
a research tool. The importance of this is deceptively simple: in astronomy,
technological innovations have often led to new discoveries. Driven by the
need to get as much observing time as possible and the desire to take advantage
of the best observing conditions, modern observatories have experimented
with new technologies and modes of collecting images and spectra. This
entailed a re-casting of the telescope by astronomers and science managers
as a factory of scientific data. At the same time, contemporary astronomers
express considerable unease and apprehension about how these technological
changes have altered, in ways subtle and profound, the nature of astronomical
observing and what it meant to be an astronomer. This short essay addresses
the issues associated with these recent changes in astronomical practice and
their connections to astronomers’ desire for ever larger and more complex
telescopes.
Keywords Telescope history · Practice of astronomy · Giant telescopes
1 Introduction
Eight years ago, I visited Cornell University. There, in the Astronomy
Department, I found a sign in a hallway that read: “Know Thyself. Know Your
Telescope.” This adage, paraphrased from Socrates, says a good deal about
W. P. McCray
Department of History, University of California, Santa Barbara, Santa Barbara, CA, USA
e-mail: [email protected]
80
Exp Astron (2009) 25:79–89
the connection between astronomers and the telescope—to do good research,
one presumably should be aware of the connection between instrument and
scientist. Albert Whitford, the director of Lick Observatory, 50 years ago in
an oral history interview called telescopes “beautiful and cantankerous instruments” and said that using them demanded what he called “high artistry.” Such
work was challenging; it could be cold, tiring, boring, or simply frustrating.
Nonetheless, time spent in the observatory dome fostered an intimate bond
between scientist and machine.
Over the last four centuries, astronomers did indeed form a personal
relation with telescopes, which are, after all, their central research tools.
However, in the last half-century, astronomers’ relationship with telescopes
has changed profoundly. These interactions have become more productive, yet
more? impersonal and remote. The classic image of an astronomer working
alone under the night sky is now mostly a romantic anachronism. Efficiency,
improved performance, comfort, and access to more data are goals that the
science community has chosen to pursue.
In studying the history of the telescope, historians have often focused on
their size and light-collecting ability. To be fair, this emphasis is understandable. In 1970, only three optical telescopes in the world had mirrors larger than
3 m. Today, sixteen telescopes have mirrors bigger than 6 m. Given the funding
and technical challenges scientists faced 30 years ago—recall that in 1978 no
one had yet demonstrated that they could even build a telescope mirror 8 m or
bigger—today’s global menagerie of telescopes seems quite remarkable, even
unexpected.
I want to, therefore, approach the last half-century of telescope history from
a perspective other than size and light collecting ability. This essay considers
how technological changes since 1945—specifically, the adoption of computers
and other electronic devices—have altered the nature of what it means to do
astronomy and to be an astronomer.
2 Historical parallels
Many changes in the astronomy community parallel earlier ones that highenergy physicists experienced, debated and worried over. In the 1950s and
1960s particle accelerators and bubble chambers grew enormously in size,
complexity, and cost as well as the amount of data generated. For many
scientists, the data itself became THE experiment. Physicists at Berkeley, for
instance, relied on machines like the Franckenstein that could automatically
measure particle tracks much faster than a human operator. As Peter Galison
has noted, experts in data-processing told physicists that they could envision
eliminating people from the data analysis process, function by function through
computerization. Berkeley physicist and future Nobel winner Luis Alvarez
defended these techniques as a necessary and pragmatic part of modern
physics, where high efficiency came from production line organization. The
Exp Astron (2009) 25:79–89
81
opinion of the physics community itself was divided over these new tools. At
CERN—where the rush to automation was even more fast-paced than in the
USA—one European physicist lamented that “in a few years. . . one would not
go to start a new experiment but would just go into the data archives, get a
few magnetic tapes, and. . . that would be the experiment. . . .This point of view
frightens me,” he said. Three decades later, many astronomers voiced similar
ambivalence.
Another historical parallel that helps us understand the changing relation
between astronomers and the telescope comes from the early days of the
manned space program. As David Mindell discusses in his recent book Digital
Apollo, hands-on flying was central to the test pilot’s professional identity,
making questions about the pilot’s role a central debate in early spaceflight
missions. As one pilot, a young man named Neil Armstrong wrote in 1960:
“this hypersonic vehicle [he was talking about the X-15] is an instrument of
the pilot” and depends on him “for control and flight success.” Armstrong’s
viewpoint conflicted with that of NASA engineers who described the Mercury
spacecraft as “an automatic system which, if all components work correctly,
can complete an unmanned flight.” The primary task of the astronaut, in their
view, was “systems management” and not actual stick and rudder flying. These
striking similarities speak not just to the increasing complexity of research tools
after 1945 but also go the heart of scientists’ professional identity.
3 Palomar nights
To get a sense of what has changed since 1948, consider how a typical night
at the 200-in. at Palomar might have unfolded six decades ago. While the
telescope operator opened the dome shutters and checked the weather, the
astronomer climbed into the prime-focus cage or down to the coudé focus.
“Each time you go up,” one long-time Palomar observer recalled, “you carry
some photographic plates in a tiny little box. . . a whole set of dreams of what
the object you’re going to work on is going to turn out to be.” The telescope
operator steered the telescope from the main control desk on the dome floor
where analog dials showed the telescope’s position. Throughout the night, the
operator pointed the telescope to where the astronomer requested through an
intercom or simply by shouting.
Many of the observation runs with the 200-in. were carried out at the telescope’s prime focus station. There researchers sat in the cramped “observing
cage” and rode with the telescope all night long while collecting data. A single
exposure might last all night. The astronomer looked through an eyepiece
almost continually to keep the object centered properly on the photographic
plate or spectrograph slit. The work could be quite tiring and even inefficient.
A sense of the personal and physical experience of using a large telescope
is revealed in an audio tape (surreptitiously made by a night assistant) of
82
Exp Astron (2009) 25:79–89
Rudolph Minkowski talking to himself during a night’s observing run on the
Palomar 200-in.:
“Where is that thing? I think that’s it over there. [sound of slow motion
motor] Damn! Wrong button. [motor again] There it is. . . now where’s
that little double to the left?. . . No, that’s not it. . . [to the night assistant]
Try a little west. . . Stop! [To himself] Here it comes [slow motion motor].
Yes, I think that’s it. Pull it down a little [slow motion motor again] Ah!
Too much! [motor sound] Now I’ve got it! [sound of camera opening;
then, to the night assistant] Start the exposure! I’ll take three hours on
this one. You can rest awhile.”
A long sigh followed as Minkowski began to hum Beethoven’s “Ode to Joy”
before the tape ended.
Using a big telescope required skill one acquired through time and perseverance. Astronomers had their own techniques for collecting data; telescope
operators even claimed they could tell who was in the observing cage simply
by the types of instructions that were being shouted out to them. But, as
Maarten Schmidt said in an interview, the astronomer had the profound,
perhaps humbling, experience of “sitting with his back to the whole universe.”
4 Was “technology supplanting artistry”?
As early as 1945, Princeton’s Henry Norris Russell predicted that “in another
decade or so, those of us who slaved looking through eyepieces. . . will be pitied
by the folks who use the new devices.” After World War Two, astronomers
gradually replaced photographic plates with electronic devices that could
capture images and spectra more efficiently and from different focal points.
These new tools slowly helped re-define the astronomer’s interaction with the
telescope. Things like auto-guiders and image intensifiers made observations
easier and more efficient. But the telescope’s complexity increased because
now astronomers had to modify all the things that they used to do by hand.
While new tools made observations more efficient, they also began to
separate the person from the machine. In some cases, this was literally the
case: television-based guiding systems brought the astronomer in from the cold
observing cage to warm, well-lit data rooms. The output of electronics systems
was displayed in a separate control room and could also be used to control the
telescope automatically.
By 1968, astronomers had managed to operate telescopes remotely. Scientists in Tucson, for instance, used punched paper tape and a computer
to remotely collect data with a small telescope on Kitt Peak. Not all astronomers appreciated such technological feats. Some experienced observers,
one scientist said, still preferred “manual controls, manual guiding, and fairly
extensive human interaction with the instruments.” At Lick Observatory,
scientist George Herbig, however, cautioned his colleagues “not [to] hold
too conservative a philosophy and be accused,” he said, “of always planning
Exp Astron (2009) 25:79–89
83
to wage the next war with the weapons that won the last.” Nonetheless,
many scientists wondered whether “technology [was] supplanting artistry” and
whether the craft of traditional observing would vanish.
One cause of astronomers’ apprehension was the rapid appearance of computers in the observatory. In the 1950s, astronomers began using computers for
routine tasks such as data reduction. These were large and costly machines centrally located on university campuses. The introduction of mini-computers, like
the IBM 360 and Raytheon’s 703 in the mid-1960s, changed this completely. As
computer equipment became more common in astronomy, astronomers saw
that it had the potential to affect other areas of science practice other than data
processing. For example, in 1966, astronomers at Northwestern University
described the use of a computerized information retrieval system that could
provide answers to technical inquiries. By processing up to eight questions a
minute, the system analyzed queries posed in English about stellar astronomy,
searched a star catalog, and provided numerical answers. Once modern observatories installed powerful yet relatively inexpensive computers, astronomers
could link them to other electronic equipment or even the telescope itself. As
this happened, observatories were obliged to hire scientists and engineers with
the skills to make the new devices work.
Edwin W. Dennison was one such person. After getting his PhD in astronomy at the University of Michigan, he was hired as a researcher at Caltech.
He soon turned his attention toward the development of electronic systems
for telescopes. In 1966, Dennison directed what came to be called the AstroElectronics Laboratory at Palomar. By 1969, Dennison’s laboratory employed
a dozen people. One of their main projects was adapting mini-computers for
direct use at the telescope. Dennison’s devices found a welcome reception at
Palomar, and by 1970 astronomers at the 200-in. used a digital data-recording
system on one out of every three observing runs.
The degree to which technological enthusiasts believed that modern electronic instruments could change astronomers’ nightly work is reflected in an
article Dennison published in Science. Entitled “Electronic Optical Astronomy: Philosophy and Practice,” Dennison’s 1971 paper outlined two principles
that guided his lab’s efforts: to enable the astronomer to make observations
that would otherwise be difficult or impossible and to improve the “operating
efficiency of the instruments.”
Besides describing the computer and electronic systems at the Hale Observatories, the article offered an overall philosophy that guided the development
of electronic instrumentation. Dennison, while recognizing that no machine
could have “the flexibility or ingenuity of the experienced observer,” expressed
concern that a fatigued and uncomfortable astronomer could accidentally
make mistakes during the night, wasting increasingly valuable observing time.
As a result, all telescope and electronic controls should be as simple and selfexplanatory as possible. Only those the observer would need during the night
should be available. The observer, according to Dennison, should be free to
monitor the data as they were collected in order to “eliminate the feeling
that he is being manipulated by the dictates of the equipment.” But, to avoid
84
Exp Astron (2009) 25:79–89
systematic errors, the observer “must never be a link in the data-collection
chain.”
5 Change we need?
Not all astronomers felt at ease with the new technologies invading the
observatory dome. At an ESO conference on telescopes in 1971, Dennison
showed a diagram of the Hale Observatories’ new computing system. On
one side of the drawing, set apart from boxes labeled “teletype” and “future
large computer” he included an icon identified simply as “Telescope.” One
astronomer, seeing this picture, remarked that in this new technology regime,
the community’s most treasured tool might just become what he sarcastically
called “a big computer with a large optical analog-input at its periphery.”
We can see the clash between these two cultures of doing astronomy at
another conference ESO and CERN jointly organized in 1974. This was an
especially critical time because astronomers throughout the world were in the
process of building the next generation of large telescopes like the AngloAustralian Telescope and the 4-m at Kitt Peak and Cerro Tololo. Debate over
how these new facilities would be used was therefore of central importance.
An entire morning session was devoted to the “Philosophy of Telescope
Use.” Speakers like Caltech’s Jesse Greenstein assured the large international
gathering that the telescope was still their central research tool. Nonetheless,
Greenstein told colleagues that, while he was impressed and comforted by
the extraordinary gains in data-gathering ability afforded by new electronic
devices, they often caused him to “return from observing in a state of personal
rage.”
The 1974 ESO-CERN meeting also featured a lively panel discussion
(chaired by Adriaan Blaauw, ESO’s director) that reflected the state of unease
and transition in the astronomy community. The astronomers who spoke that
spring morning in Geneva reflected two different visions for the future of large
telescope use. Harry van der Laan, a future ESO director, suggested that it
might be better if the astronomer “doesn’t go near the [observatory]. . . he
communicates by telephone and telex and gets his data back.” Of course there
would be exceptional cases where interaction between the astronomer and the
observatory staff might be necessary, but “big modern optical telescopes are
evolving. . . where it should be mostly a hands-off policy as far as the using
astronomer is concerned.”
Another ESO astronomer, Jaap Tinbergen, made even more pointed remarks, saying that optical astronomers were “just too conservative, and too
lazy in some cases, to try it any other way.” He then proceeded to tell a story
about a neophyte astronomer who wasted a night’s worth of observing because
he had not properly aligned his target on the spectrograph’s slit. “Now that
wouldn’t have happened,” Tinbergen concluded, “if he had allowed his spectra
to be taken by an assistant.”
Exp Astron (2009) 25:79–89
85
Greenstein reacted angrily to this debate. He warned his colleagues not to
make the mistake of believing that “they will be able to send messages” over
the phone to the observatory and get reliable data. “I wouldn’t trust anybody
else,” he contended, “Either you believe that the ‘Establishment’ has been a
bunch of idiots. . . who like to be up on a freezing mountain, or you must believe
that there was some point in it.”
For tradition-minded astronomers like Greenstein, who had worked at
telescopes of all sizes for nearly 40 years, the idea that the astronomer
might not directly participate in data collection was anathema. For other
astronomers, many of whom lacked California scientists’ unparalleled access
to big telescopes, there was not nearly enough telescope time to satisfy the
needs of the community and any new technology that made observing more
efficient was worth considering.
6 Reconsidering telescope design
Despite the reservations of Greenstein and his colleagues, these computer
tools were incorporated directly into the design and construction of new and
existing telescopes. Previously, large telescopes were understood by engineers
as a system of separate parts—mirror, truss, dome. Yes, they interacted but
their design was largely done independently. By the 1970s, telescope engineers
started using computers to assist with engineering analyses. Prior to this,
computer-aided design and analysis programs were not widely used, and telescope design was as much art as science. But, by separating the telescope into
different “analytical segments,” computer-assisted modeling, finite element
analysis, and CAD programs let engineers model the mechanical behavior
of telescopes before construction. This prompted one astronomer in the early
1970s to remark, “Now we have got rid of the observer, do we get rid of the
engineer too by putting everything into a computer?”
Of course, as computer and electronic systems for the telescope became
more sophisticated, there were changes in how the telescope was actually operated. For example, in the 1970s, project engineers working on the sophisticated
Multiple Mirror Telescope (MMT) on Mt. Hopkins in southern Arizona built
a facility that required a complex and connected system of lasers, electromechanical servos, and optical sensors.
Unlike conventional telescopes of the time, the MMT had no observer cage
where the astronomer could ride and guide the telescope while collecting data.
From the beginning, the MMT’s designers envisioned that astronomers would
operate the telescope remotely from a control room and move the telescope
using a television system while monitoring the output of data on a computer
display.
The sophisticated optics system of the MMT, for example, carefully combined images from each of the six 1.8-m primary mirrors and depended on
three different electronic systems, linked together, to manage the movement
86
Exp Astron (2009) 25:79–89
of the telescope, the co-alignment of the mirrors, and its scientific instruments.
Its ability to point and move quickly aided its efficient use. The astronomer was
indeed becoming much more of a systems manager. MMT engineers actually
went so far as to describe the telescope itself as a “large and unique peripheral
device” for a computer system.
7 New modes of practice
Since 1945, scientists and engineers transformed the large optical telescope
into the most visible part of a much bigger astronomical data-collecting
network. Modern observatories like Gemini and the VLT are tools in a
transoceanic, decentralized, data-oriented and collaborative research environment. They are linked in real-time to engineers and astronomers by high-speed
data networks, fiber optic cables, video conferences, and the World Wide
Web. This shift has entailed a re-casting of the telescope by astronomers and
science managers as a factory of scientific data and scientists as customers who
order up astronomical data that is delivered to them electronically while they
monitor the process through Internet links.
Building giant new telescopes like Gemini and the VLT involved more
than manipulating glass and steel to construct complex scientific instruments.
For centuries, using a telescope meant going to the observatory at night,
often alone, and collecting data. The advantage of classical observing was
that control of the telescope rested with the astronomers. If they desired,
scientists could always point the telescope at someplace new and hope for a
discovery based on serendipity and instinct rather than extensive pre-planning.
According to Alan Dressler, an astronomer at the Carnegie Institution of
Washington, “A certain degree of cowboy-like behavior made a difference,”
Dressler explained, “People made their reputations based on how they performed at the telescope.”
Ground-based astronomers who observed in the classical style, of course,
took risks. If the weather was poor or cloudy they would collect few data. If
an astronomer had an important observing program that called for the best
possible seeing conditions and these did not materialize, then he had to fall
back on other programs for that night. Classical observing was, in other words,
something of a gamble when it came to getting data. There were more serious
risks as well. In 1987, Marc Aaronson, a promising young scientist and skilled
observer, was fatally injured while using the 4-m telescope at Kitt Peak.
As the new generation of giant telescopes saw first light in the last 15 years or
so, astronomers developed new strategies for doing science with them. Many
of the world’s modern observatories, for instance, use some form of “queue
observing.” Adapted from the space telescope world, this mode enables observatory staff to collect scientific information and provide it in a standard format
to researchers. The same staff uses computer algorithms to control scheduling
and the order of executing research projects, taking into consideration weather
conditions and the facility’s time allocation agreements.
Exp Astron (2009) 25:79–89
87
Astronomers’ traditional concept of an “observatory” came to embrace a
panoply of interdependent elements—the telescope’s enclosure, instruments,
mirror supports and control systems, data archives, and electronic linkages
for remote observing. Tying these myriad systems together at Gemini or the
VLT were miles of cables, wires, and fibers. Further complexity was added
by the integration of new tools such as videoconferencing and Internet access
which would allow astronomers to monitor the telescopes and participate in
data collection remotely. Even local weather conditions on Mauna Kea and
Cerro Pachón constituted part of the “observatory system.” To optimize the
thermal control of a thin primary mirror, for example, engineers needed to
keep it close to the predicted temperature of the night air. Successfully doing
this meant that staff needed extensive and accurate information on how the
weather on the mountains varied seasonally as well as a system for predicting
nighttime observing conditions from real-time meteorological data.
Advocates of queue observing insisted it enhanced efficiency. Unless they
were familiar with the telescope and its instrumentation, astronomers might
waste several hours in calibrating equipment before collecting any data. When
the Very Large Telescope was completed in 2000, ESO managers had thousands of hours of telescope time to allocate annually. Each telescope cost
about a dollar per second around the clock. “If a staff member’s work can
save two nights,” one European scientist noted, “their salary is already accounted for.” In many ways, observatories’ emphasis on flexibility, adaptation,
and streamlined efficiency resembled Japanese “just in time” manufacturing
practices admired by American and European business leaders in the 1980s.
This style of research challenged some traditional views of what being an
astronomer was about. Like Alan Dressler, Matt Mountain (former director
of the Gemini Observatory and now head of the Space Telescope Science
Institute) also opted for a cowboy analogy. “I do not believe discoveries are
made at the telescope riding the weather like a cowboy riding a bucking
bronco,” he claimed, “Discoveries emerge at a computer terminal while trying
to reconcile an awkward data set with preconceived models.”
8 Data and more data
One of the things that strikes me when I read plans for the next generation
of giant telescopes and sky surveys is the attention given to data flow and
management—“science-oriented systems engineering.” Just as Dennison located the telescope on the periphery of his chart in 1971, today’s astronomer
has become one node in the data collection chain, their expertise augmented
by a whole host of specialists and systems. The data pipeline has become a
central organizing idea in designing new telescopes and the telescope itself
was transformed, like the atom smashers of the 1950s and 1960s, into a data
factory. Astronomers’ new lexicon included “productivity,” “efficiency,” and
“mission statement,” terms more common in the corporate boardroom than
the telescope dome. Observatory managers began to speak of their colleagues
88
Exp Astron (2009) 25:79–89
as “customers” who actively shopped for telescopes equipped with the right
instruments to make particular measurements.
Determining the catalyst for this shift in astronomers’ thinking is difficult.
The scientists themselves attributed the changes primarily to technological progress—bigger mirrors, more ambitious telescopes, and much greater
computing power. Since the 1970s, the amount of information collected by
telescopes worldwide had far outpaced gains made in light collecting power
alone. While the total area of telescope mirrors around the world had increased
by a factor of 30 or more since 1975, the number of pixels in a typical CCD
detector grew by a factor of at least 3,000. The introduction and maturation of
new communications tools—the World Wide Web, the Internet, videoconferencing, and high-bandwidth links connecting mountain peaks in Hawaii and
Chile to astronomers in the mainland USA—also motivated astronomers to
adopt new patterns of work and consider changes in their nighttime use of
telescopes. Moreover, well-publicized science results from the Hubble Space
Telescope encouraged scientists to maximize the capability of their new tools.
Not all changes occurred as a result of technological innovations. Additional
impetus came from observatory directors and science managers who argued
that the cost of building and operating new telescopes demanded greater
efficiency. Some in the community recognized that even with smaller (and supposedly less complex) telescopes, the first night astronomers spent observing
was often unproductive. In today’s modern observatories, software engineers
and observatory directors give considerable attention to the most efficient
manner in which to examine, manage, and archive the vast amounts of data
produced. To date, the Hubble Telescope has amassed some 120 terabytes
of data. Future survey facilities like the Large Synoptic Survey Telescope are
expected to produce that amount of data every week, a glut that has prompted
some science writers recently to wonder whether the flood of data will presage
an era where theory is less important, even unnecessary. Is today’s astronomer
someone who spends time at the telescope or has this occupation shifted to
become much more focused on data management and mining?
9 Conclusion
Think back to that sign at Cornell that I described at the start of this essay.
What does the injunction “know your telescope” mean in such a data-centric
universe?
To some astronomers, first light at new observatories like the Very Large
Telescope represented the loss of the romance they associated with telescopic
observing and with it of? a certain degree of control. But with this loss came
revitalized powers to observe the universe more efficiently.
Over the past half-century, a new generation of eyes –flesh and blood, glass
and steel—has turned to the cosmos. If we look more closely at the practice of
astronomy and how telescope technology has affected it, it seems to me that
these people and their instruments increasingly appear as cyborgs—a blend
Exp Astron (2009) 25:79–89
89
of person, machine, computer working together. Since 1945, the telescope has
merged with the computer, the web, the database into a hybrid instrument.
But there was nothing inevitable about this change. Computers, chips,
circuits, and databases have not re-made astronomy. Rather, astronomers
chose these new tools and systems and integrated them into their nightly work
routines. And while scientists use them in a manner radically different from
yesteryear’s artistry, telescopes—still beautiful and still cantankerous—still
produce new and exciting knowledge about the universe.
Bibliographic note For this paper, I drew upon two books that were helpful for setting up
comparative examples; these included several chapters in Peter Galison, Image and Logic: A
Material Culture of Microphysics (Chicago: University of Chicago Press, 1997) and David A.
Mindell’s Digital Apollo: Human and Machine in Spaceflight (Cambridge, MA: The MIT Press,
2008). Other than the discussion of particle physics and test pilots (which come from Galison and
Mindell’s books), all direct quotes are taken from my 2004 book Giant Telescopes: Astronomical
Ambition and the Promise of Technology (Cambridge, MA: Harvard University Press, 2004);
readers interested in the original sources of quotes used are directed to it.