. 4
( 28)


sufficient number of electrons are moving for a
current to be detected. In the past, these photo- the ratio of the signal to the uncertainty, some-
multiplier devices allowed only a single detection times called the signal-to-noise ratio, is N. So, as
you increase N, the signal-to-noise ratio increases
element in the focal plane. That is, there is one
but only as N. For example, if you want to
pixel. So, they provided higher efficiency, but at
the cost of the panoramic quality. improve the signal-to-noise ratio by a factor of
One problem with any detector is that it pro- two, you would have to increase the number of
duces some background level. This is sometimes counts by a factor of four.
called the dark current, because it is present even More recently, astronomy has been revolu-
tionized by the development of charge-coupled-
when no light is shining on the detector. This
devices (CCDs). They provide a grid of detectors all
generally results from thermal emission from the
detector. For most astronomical observations, with high quantum efficiency (greater than 50%
this background is much stronger than the signal and sometimes as high as 80% to 90%). Each ele-
you are trying to detect. The background has two ment of the grid is one pixel, and keeps an elec-
deleterious effects. The first is that it must be sub- tronic record of the intensity of light striking its
tracted from any measurement, to just give the position. The array is on a single silicon chip.
value of the astronomical signal. The second is There are typically over 1 million pixels (1000
that it produces a random fluctuation in the 1000) on astronomical CCDs. (Commercial digital
measurement. The stronger the background, the cameras use CCDs.) Each pixel is a potential well
higher the fluctuation. This fluctuation produces that traps electrons. For the most part, the elec-
an uncertainty (statistical error) in any result. trons are the result of photons striking the pixel.
To see how this effect works, it is easiest to In order to use the information on the chip, there
think in terms of numbers of counts in a given has to be some way of reading it into a computer.
measurement. For a photomultiplier, the num- This is illustrated in Fig. 4.18.
ber of counts in the signal is the number of pho- An advantage to a CCD is that it is nearly lin-
tons striking the detector, multiplied by the ear in its response. This means that the number of
quantum efficiency. For the background, we can electrons is proportional to the number of strik-
think of the number of photons that would be ing photons. This is true over a wide range of
equivalent to the emission from the background. intensities. As with photomultipliers, there is
If the background is thermal emission, that some dark current. This can be reduced by cooling
number would be kT/h , where T is the tempera- the detector. The dark current is generally meas-
ture of your background emission and is the ured by taking exposures with no light entering.
frequency at which you are measuring. From There is also a variation in sensitivity from pixel
this, you see that we can reduce the background to pixel. This can be measured by making an expo-
by cooling the detector. sure of a uniform field, such as the twilight sky or
the inside of the dome. This process is called flat
The effect of this fluctuation is to cause a
fielding. CCDs are so stable that dark current meas-
scatter in the results of counting experiments. If
you want to determine the average rate at which urements and flat field measurements only need
you are counting photons from a given source, to be done a few times a night.
you might measure for some time, say ten sec- If a cosmic ray (cosmic rays are charged parti-
onds. If you repeat this experiment many times, cles that permeate interstellar space) strikes a
you will find that the number of counts is not CCD during an exposure, it will make a few pixels
always the same. If you plot a histogram of the look very bright. These can easily be removed in
number of times each result comes out, you will the computer processing of the image. To mini-
find a gaussian centered on some value. That mize their effects, it is better to take a few short

range, meaning that we can see faint objects in
0 D1
A1 B1 C1 D1 A1 B1 C1
the presence of bright ones.
0 A2 B2 C2 D2
A2 B2 C2 D2
When photometric observations are being
A3 B3 C3 D3 A3 B3 C3
0 0 D3
made, we generally compare the brightness of the
A4 B4 C4 D4 0 A4 B4 C4 D4
star under study with the brightnesses of stars
whose properties have already been studied. By
changing filters we can measure, for example,
Comp Comp

the U, B and V magnitudes of a star, one after
(a) (b)
another. Some method of recording the data is
still needed. One option is photographic. The
0 0
0 A1 B1 C1
0 A1 B1 C1
brighter the star is, the larger its image on a pho-
D1 0
A2 B2 C2
A2 B2 C2
tographic plate. (This is an artifact of the photo-
A3 B3 C3
D2 0
A3 B3 C3 0
graphic process and atmospheric seeing.) We can
0 0
A4 B4 C4 A4 B4 C4
measure the brightness of a star by measuring
the size of its image. (Remember the actual
Comp extent of the star is too small to detect in our
(c) images.) Photoelectric devices are well suited for
photometry. Almost all photometry is now done
using photomultipliers or CCDs. Some of the
0 A1 B1
standard colors even account for the wavelength
A2 B2
responses of various commercially available pho-
A3 B3 C3
0 A4 B4 C4

4.5.2 Spectroscopy
In spectroscopy we need a means of bringing the
(e) image in different wavelengths to different phys-
ical locations on our detector. We have already
Fig 4.18. How a CCD readout works. In this example, we
seen that this can be done with a prism. Since a
just look at 16 pixels, with rows numbered and columns let-
tered. The readout device is on the right. (a) An exposure is prism does not spread the light out very much,
¬nished. Each pixel has a value indicated by the label for that we say that the prism is a low dispersion instru-
pixel. The readout device has levels zero. (b) The numbers ment. Dispersion is a measure of the degree to
are all shifted one pixel to the right. The ¬rst row is now
which the spectrum is spread out. Low dispersion
zeroed, and the readout device has the contents of the last
spectra are sometimes adequate for determining
row. (c) The contents of the readout row are shifted down
the spectral type of a star. Sometimes a thin
one, as the ¬rst number (bottom right value) is read. (d) The
prism is placed over the objective of the telescope
process continues until that whole row is read. (e) Everything
and a photograph is taken of the whole field.
shifts one more to the right, and the readout process of the
Instead of seeing the individual stars, the spec-
last row repeats. This then continues until all values have
been read and pixels have been set to zero. trum of each star appears in its place. These objec-
tive prism spectra are quite useful for classifying
large numbers of stars very quickly.
exposures and add them together (using a com- When better resolution is needed, we gener-
puter) than one long exposure. There is also an ally use a diffraction grating, illustrated in Fig. 4.19.
error introduced in the readout process, called For any wavelength , the grating produces a max-
readout noise. This can put a limit on the faintest imum at an angle given by
signals CCDs can see.
d sin m (4.4)
Having the image in computer readable form
is actually very convenient, because many new where d is the separation between the slits and m
techniques are being used to computer enhance is an integer, called the order of the maximum.
very faint images. This provides a large dynamic The higher the order is, the more spread out the




Light Interference







Fig 4.19. Diffraction grating. Light comes in from the upper
left.The beam re¬‚ected off each step spreads out due to dif-
fraction. However, interference effects result in maxima in the
indicated directions.The angles of the steps can be adjusted
(blazed) to throw most of the light into the desired order. Constructive

Fig 4.20. Operation of an interference ¬lter.

spectrum. Suppose our grating just lets us sepa-
time, and must keep changing the spacing, d, to
rate (resolve) two spectral lines that are apart
obtain a complete spectrum. Another problem is
in wavelength. The resolving power of the grating is
that different orders (m) of different wavelengths
then defined as
can get through at the same time. You can solve
R / (4.5)
this problem by adding a second filter with a dif-
ferent spacing, set to pass the desired wavelength
If the grating has N lines in it, then in the
and remove the unwanted orders. A device with
order m the resolving power is given by
multiple interference filters is called a Fabry“Perot
R Nm (4.6)
A major recent improvement has been the
Some gratings have over 10 000 lines per cen-
development of devices that produce a Fourier
timeter over a length of several centimeters.
This means that resolving powers of 105 can be transform of the spectrum. These devices provide
astronomers with a great deal of flexibility and
achieved. In general, light will go out into sev-
sensitivity. Fig. 4.21 shows the operation of one
eral orders. It is possible to cut the lines of a
such device, called a Michelson interferometer. The
grating so that most of the light goes into a par-
incoming radiation is split into two beams, which
ticular order. This process is called blazing.
are reflected off mirrors so that they come back
It is possible to use interference filters such as
to the same location and interfere with each
that shown in Fig. 4.20. There are two flat parallel
other. The path length of one of the beams can be
reflecting surfaces placed close to each other.
altered by moving a mirror. This changes the
There is a maximum in the transmitted radiation
phase of the incoming beams. By seeing how the
when twice the spacing between the surfaces, d,
intensity changes as we move the mirror, we form
is equal to an integral number of wavelengths.
an idea of the relative importance of longer and
That is
shorter wavelength radiation.
2d (4.7)
According to Fig. 4.21, the total path length
difference is x. We look at the electric field for
One problem with this approach is that we
each wave. In this case it is convenient to write
can only measure a small wavelength range at a

Mirror be the total power, and we define I( k) I(k), this
simplifies to

I1k 2exp3 ikx4 dk
I1x2 2I0 (4.11)

The integral in this expression is the Fourier

transform of I(x). So by measuring I(x), we are also
Incoming D + x/2
measuring the Fourier transform . This means
that we can find I(k), power as a function of wave-
Beam length, from the inverse Fourier transfrom of I(x),
Splitter which is
I1x2exp3 ikx4 dk
Detector q

In a real measurement, we don™t measure I(x)
Fig 4.21. Michelson interferometer. Light enters from the
left, and strikes a beam splitter.The split beams bounce off for all values of x. There are two limitations. One
mirrors and are brought back together to interfere with is the total range over which we move the mirror.
each other. One of the mirrors has its position ¬xed, and the
This limits our ability to do the integral from
other is movable.
minus to plus infinity. The other is that we can
only move the mirror in finite steps. This means
that we only measure I(x) at those positions, so
the waves as E0 exp[i(kx t)], where k 2 / and
this limits our ability to approximate the inverse
2 , and E0 is the electric field amplitude of
transform integral as a sum. The closer together
the wave. So, the two waves that will be recom-
we measure I(x), the shorter wavelengths (higher
bined can be written as
frequencies) we are sensitive to. The greater the
E1 E0 exp[ i t] (4.8a) largest value of x at which we measure I(x), the
more information we have on the longer wave-
E2 E0 exp[i(kx t)] (4.8b)
lengths (lower frequencies), so this sets the limit
Taking the total electric field, E E1 E2, and in the frequency resolution in the computed
the intensity, I EE*, we can write (see Problem spectrum.

2E2 c 1 d
exp1ikx2 exp1 ikx2
4.6 Observing in the ultraviolet
I1k, x2 (4.9)

For any position of the mirror, corresponding to a
The visible part of the spectrum only gives us
path length x, we will receive the contributions
access to a small fraction of the radiation given
from all wavelengths (k). So, to find the total
off by astronomical objects. For centuries, how-
intensity as a function of x, we integrate over all k:
ever, this was the only information available to
astronomers. We will see throughout this book
I1x2 I1k, x2dk
that observations in other parts of the spectrum
have revealed entirely new types of objects or
2 I1k 2 c 1
d dk
exp1ikx2 exp1 ikx2
provided us with information crucial to under-
standing objects that are already observed in the
2 I1k 2 dk I1k 2 exp1ikx2 dk I1k 2 exp1 ikx2dk
q q q
visible. In discussing other parts of the spectrum,
we start with ultraviolet observations, because
0 0 0
the techniques are very similar to those in optical
If we let
I1k 2dk In many ways, we can think of ultraviolet
I0 (4.10)
observations as being short wavelength visible

observations. The basic imaging ideas are the at 10 m. This is not a problem for optical detec-
same. Of course, since the wavelength is shorter, tors, but it is a problem for infrared detectors.
mirror surfaces must be more accurate in the (See Problem 4.23.) In an infrared telescope, the
ultraviolet than in the visible. The normal coat- radiation paths must be carefully designed so
ings that we use to make mirrors reflective in the that the detector cannot ˜see™ any hot surface.
visible do not work as well in the ultraviolet, and Some reduction in the problem can be obtained
different coatings are needed. Since ultraviolet by cooling surfaces that can radiate into the de-
photons have more energy than visible photons, tector. Such infrared optimization techniques are
the uv photons can easily be detected with pho- being incorporated into the northern hemisphere
tographic plates, photomultipliers or CCDs. (Mauna Kea) part of Project Gemini (Fig. 4.14).
The major problem is that ultraviolet radia- Detectors used in the infrared are generally
tion does not penetrate the Earth™s atmosphere. different from those used in the visible. Infrared
If you don™t go too far into the ultraviolet, some photons are not energetic enough to expose nor-
observations are possible at high altitudes. mal photographic emulsion. Recently, infrared
However, we have become increasingly dependent sensitive emulsions have been developed.
on ultraviolet satellites. Some pioneering satellites Infrared photons also have a hard time causing
were Copernicus (1972“1981) and International electrons to be ejected from metals. One of the
Ultraviolet Explorer (IUE, 1978“1996). IUE had a great advances of the past few years has been the
0.45 m mirror, 3 arc sec angular resolutions and development of efficient infrared arrays of detec-
an R 12 000 spectrograph. Currently, we can use tors. They read voltage, rather than the current in
HST, whose mirror was designed to work in the a CCD, but have readout schemes similar to CCDs.
ultraviolet as well as the visible. Far Ultraviolet Infrared arrays also have a smaller number of pix-
Spectroscopic Explorer (FUSE) was launched in els, typically 32 32. The detectors are cooled to
1998, with a 0.64 m mirror and a high resolution reduce background noise. (As an aside, these
spectrograph. arrays were first developed by the military to put
into satellites looking down at the Earth. With
the end of the cold war, this technology became
4.7 Observing in the infrared declassified.)
Originally, the most common type of infrared
In this section we briefly look at some of the detector was called a bolometer. A bolometer is a
techniques for observing in the infrared part of device that heats up in a known way when radia-
the spectrum. For some purposes we can simply tion falls on it. We generally use a material whose
think of infrared radiation as being long wave- electrical properties change with temperature. For
length visible radiation. In fact, much of infrared example, if the resistance of a bolometer changes
astronomy is done on normal optical telescopes. with temperature, we can measure temperature
The long wavelength means that surface accu- changes by measuring resistance changes. By
racy of mirrors is not a problem. A surface accu- measuring the temperature increase, we can
rate enough for optical observations is certainly determine the total amount of energy striking the
accurate enough for infrared observations. bolometer. (Remember, in Chapter 2, we defined a
However, the longer wavelength makes diffrac- bolometric magnitude based on the total amount
tion more of a problem. For example, for a 1 m of energy given off by a star.)
diameter telescope working at a wavelength of Spectroscopy in the infrared is different than
10 m, the diffraction limit is 2 arc sec, slightly in the visible. One problem is that the longer
worse than the seeing limit at a good site. wavelength means that objects must be physically
One problem with infrared observations not larger to provide the same spectral resolution.
common with optical observations is radiation Another problem is the thermal emission from the
from the telescope itself. Parts of the telescope material used to construct the devices. If the sur-
that are not perfectly reflective radiate like black- faces can be cooled, then this problem is reduced.
bodies at temperatures close to 300 K, with a peak Prisms are of some value for low resolution. Cooled

gratings are used. It is possible to use tunable inter- We call the wavelengths at which some obser-
vations are possible from the ground infrared win-
ference filters (Fabry“Perot interferometers) or
dows in the atmosphere. Fig. 4.22 shows some of
Michelson interferometers, as discussed earlier.
The major problem in the infrared is the the major infrared windows. At the very least, the
Earth™s atmosphere. The atmosphere is totally 2 km altitude of many optical observatories is
opaque at some infrared wavelengths, and is, at required. In general, 2 km is only sufficient for
best, only partially transparent at all other working in the near infrared, at wavelengths of a
infrared wavelengths. The opacity of the atmos- few micrometers. If we want to work farther into
phere causes two problems: (1) the atmosphere the infrared, higher altitudes are necessary. Some
blocks the infrared radiation from the sources we observatories have been placed at altitudes as
are studying; (2) the atmosphere emits its own high as 4.3 km (even though higher elevations
infrared radiation, which can be much stronger result in difficult working conditions). For exam-
than that received from the astronomical objects. ple, there are a number of infrared telescopes on
To observe with this atmospheric emission it is Mauna Kea (Fig. 4.13b).
generally necessary to compare the astronomical For many studies, even higher altitudes are
needed. For 20 years NASA operated the Kuiper
source you are looking at with some empty sky
Airborne Observatory (KAO). The KAO was a con-
nearby, thereby canceling the effects of the
atmosphere. However, this limits you to studying verted military transport (a C141), that carries a
relatively small sources. 0.9 m infrared optimized telescope to altitudes
up to 45 000 ft, for 7 hour observing sessions. The
KAO was operated as a national facility, with qual-
ified astronomers submitting proposals for
observing time. It made approximately 80 flights
per year. It operated out of the NASA Ames

Research Center (Moffet Field, CA), but could

change its base when the astronomical need dic-
tated. For example, there were regular observing
sessions in the southern hemisphere from
Christchurch, NZ. There were also customized

flights to look at transient astrophysical phenom-
1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6
Wavelength ( m)
ena such as solar eclipses. The KAO was taken out
of service in 1997.
To replace KAO, NASA is building, in coopera-
tion with astronomers in Germany, the Stratospheric
Observatory for Infrared Astronomy (SOFIA), shown in

Fig. 4.23. To allow for a larger (2.6 m) telescope it

will be made from a converted Boeing 747 SP. This
aircraft is also capable of cruising at higher alti-
tudes and providing for longer (up to 16 hour)
0.2 1.0 mm H2O
flights. For even higher altitude work, balloons are
3.0 mm H2O
used up to 100 000 ft.
6 8 10 12 14
For some observations, even a minimal
Wavelength ( m)
atmosphere causes problems, and we carry out
observations from space. Fig. 4.24 shows two of
Fig 4.22. Infrared windows.Transmission as a function of
the primary infrared space missions. One of the
wavelength is shown for an observatory at sea level. At low
altitudes good transmission (close to unity) is in only a few important early missions was the Infrared
narrow wavelength ranges, or windows. As one goes to Astronomy Satellite (IRAS), a joint American“
higher observing altitudes, the windows cover a wider range
Dutch“British project launched in January 1983.
of wavelengths. [NOAO/AURA/NSF]
The 0.6 m diameter cooled telescope was primarily

Fig 4.23. Artist™s impression of the Stratospheric
Observatory for Infrared Astronomy (SOFIA), which will be
made from a converted 747 SP. [NASA]
IRAS™s lifetime was devoted to a systematic survey
of the sky. A certain fraction of the time was
devoted to specific objects. The large scale survey
revealed over 100 000 point sources and a network
designed for imaging observations. It contained
of extended infrared emission. The whole set of
arrays of detectors operating in four wavelength
data is available as a resource to the general astro-
ranges, centered roughly at 12, 25, 60 and 100
nomical community. An astronomer interested in
m. From its high (560 mile) polar orbit, much of

Fig 4.24. Infrared satellites.
(a) Artist™s conception of Infrared
Space Observatory (ISO)


Box 4.1. Methods of displaying images.

When we look at a normal optical photograph of some
astronomical object, we have a sense of how our brains
should interpret that image. In a sense it is how the
object would look if we could view it through a large tel-
escope, or if we could somehow be transported close
enough to the object so we could see it with this detail
with the unaided eye. However, what does it mean when
we display a radio image like that in Fig. 4.25?
There is even a terrestrial analogy to this question.
You have seen ˜night vision glasses™, which allow you to
˜see™ even with no illuminating light. Remember, we nor-
mally see earthbound objects as they re¬‚ect sunlight or
roomlight, our eyes being sensitive to the range of wave-
lengths at which the Sun™s emission is strongest.The night
vision glasses work differently: they actually detect the
infrared radiation given off by objects (most of which are
usually close to 300 K). So, the night vision glasses have
infrared detectors, but our eyes are not sensitive to that
infrared radiation.Therefore the glasses also convert that
infrared image into an optical image, usually with the
brightest part of the optical image corresponding to the
strongest infrared emission. So, the image you see is an
optical representation of the infrared image.
We can do the same thing with astronomical infrared
(or ultraviolet, radio, etc.) images. We can make a false
Fig 4.24. (Continued) (b) Space Infrared Telescope Facility
gray-scale image, by creating an optical image where
(SIRTF) in the laboratory for testing. [(a) ESA/ISO; (b) NASA]
brighter regions correspond to stronger infrared emission.
It is important to remember that, while such images are
often constructed to have a true-looking appearance, they
are just a particular representation for that image.
a particular object can check the data at the
Sometimes our eyes are better at picking information out
Infrared Processing and Analysis Center (IPAC), at
of a color image than a black and white (or gray) image.
Caltech, which is not too far from the Jet
It is therefore sometimes useful to make false color images.
Propulsion Laboratory (JPL). IPAC has also become
In this case, we arbitrarily assign a color to each level of
the curator of other infrared data.
infrared emission. Often the colors will run through the
More recently, astronomers have been able to
spectrum from red to blue (or the other way). Often, a
utilize the Infrared Space Observatory (ISO), Fig.
sample bar will be placed next to the image, showing
4.24(a), a project of the European Space Agency.
what intensity level each color represents.
ISO provides a wavelength coverage extending
So far, we have been talking about what we do when
farther into the infrared, a larger telescope,
we have one piece of data at each location, say the aver-
arrays of more sensitive detectors for good imag-
age intensity in some particular wavelength band. Suppose
ing, and for the first time the ability to make
we have observed in more than one band (for example,
high quality spectra. During the HST servicing
IRAS observed in four infrared bands).We could certainly
mission, NASA added an infrared camera and
make a separate false gray or color image of each band
spectrometer (NICMOS). NASA is now making
(and we often do this), but what if we want to compare
plans for the Space Infrared Telescope Facility (SIRTF),
the bands, or simply display all the information together?
Fig. 4.24(b).


Fig 4.25. Various representations of a radio image. In this
band blue (mimicking what happens in the visible part of
case it is a 6 cm wavelength image of the Orion Nebula (which
the spectrum).We then have a false color image in which
we will discuss in Chapter 15), made with the Byrd GBT
the color has some intrinsic meaning (in that hotter
(which we will discuss in the Section 4.8). (a) Contour map.
(b) Gray-scale map. (c) False color image. (d) Color contours, in objects will appear bluer).
which colors change where there would be a contour line. We should point out that this technique can also be
[D. Shephard, R. Maddalena, J. McMullin, NRAO/AUI/NSF] used to make a true color visible image.You might say that
if you want a color image you simply use color ¬lm and
We then make a false gray-scale image of each band. We take a picture. However, no two types of color ¬lm are the
then tint each band a different color, generally making the same. Some are meant to enhance skin tones and are set
longest wavelength band red and the shortest wavelength to emphasize reds, for example. Therefore the way to

the galactic center. It was realized that astronom-
make a color photograph that looks like what you would
ical objects can be strong radio sources.
see with your eyes, we take a series of black and white
The discovery was not followed up immedi-
images, through red, green and blue ¬lters.We then com-
ately. In fact, for a long time there was only one
bine the images, utilizing the various wavelength ranges in
active radio astronomer. Grote Reber was an ama-
the same proportion as the eye uses them.This technique
teur radio astronomer in Illinois, who carried out
is well suited to making ˜true color™ images with CCDs.
observations on his back yard radio telescope in
There is another method of displaying two-dimensional
the 1930s and early 1940s. (When Reber submit-
images, contour maps. You should be familiar with topo-
ted his first paper for publication in The
logical maps on Earth, which are normally displayed as
Astrophysical Journal, it was sent to a referee, a
contour maps. All points within a given contour level have
normal procedure. To make sure that the data
a value (e.g. average intensity in a particular wavelength
were to be believed, the referee, Bart Bok, a Dutch
range) greater than the value assigned to that level.
astronomer, then living in the US, took the
Contour maps give a good feel for how the quantity you
abnormal step of visiting Reber and his telescope,
are displaying changes over some region. The closer
and taking the editor along. Bok recommended
together the contours, the more rapid the variation in
publication of the paper, and was the first tradi-
the plotted quantity.
tional optical astronomer to understand the
importance of radio astronomy. Following WW II,
radio astronomers benefitted from the develop-
ment of radar equipment during the war. Radio
4.8 Radio astronomy observations were pursued by the British, Dutch,
Australians, and a small group of Americans at
Radio observations provide us with very different Harvard. A major advancement was the ability to
information from optical observations and use observe spectral lines in the radio part of the spec-
very different techniques. The long wavelength trum. We will discuss these lines in Chapter 14.
means that the wave nature of the radiation is By the mid-1950s, it was clear that a major
very apparent in the observations. The long wave- radio observatory had to be a cooperative effort,
length also corresponds to low energy photons. and the National Radio Astronomy Observatory
This means that radio regions can tell us about (NRAO) was founded. (This was the first US
cool regions. For example, we will see how radio national observatory, being formed a little before
observations tell us about star formation in Part the optical observatory on Kitt Peak.) Bart Bok
IV. We will also see that there are high energy played a major role in the founding of the NRAO.
sources that give off much of their energy at The first telescopes of the NRAO were in Green
longer wavelengths. Thus, radio observations also Bank, West Virginia, far away from sources of
give us a way of studying high energy phenomena. man made interference (in the National Radio
Radio astronomy owes its origins to an acci- Quiet Zone). Since the Earth™s atmosphere is vir-
dental discovery by Karl Jansky, an engineer at the tually transparent through much of the radio
Bell Telephone Laboratories in New Jersey. In part of the spectrum, it is not necessary to place
1931, Jansky detected a mysterious source of radio radio observatories at high altitudes or clear
interference. He noticed that this interference sites. We can even observe through clouds. We
reached its peak four minutes earlier each day. can also observe day or night, since the sky does
This timing suggests an object that is fixed with not scatter radio waves from the Sun the way it
respect to the stars. (This four minute per day scatters light from the Sun, making the sky
shift is caused by the Earth™s motion around the appear bright (blue).
Sun. This and other aspects of astronomical time- We now take a look at how a radio telescope
keeping are discussed in Appendix G.) The time of works. A radio telescope consists of some element
maximum interference coincided with the galac- that collects the radiation and a receiver to detect
tic center crossing the local meridian. Jansky con- the radiation. Most modern radio telescopes have
cluded that he was receiving radio waves from a large dish to collect the radiation and send it to

of smaller panels that are easier to machine accu-
rately. The panels are then aligned to produce the
best surface. The alignment is at least adjusted
Radio for the effects of gravity as the telescope tilts at
different angles, and techniques are being devel-
oped to control the surface actively by monitor-
ing the panels at all times during observations.
The best resolution for single radio telescopes is
about 30 arc sec, slightly better than the naked
∆θ ∆θ ∼ »/D
eye for visible viewing.
Example 4.5 Strength of radio sources
We measure the strength of radio sources in a unit
called a Jansky (Jy). It is defined as 10 26 W/m2/Hz
D reaching our telescope. For a 1 Jy source, calculate
the power received by a perfect antenna with an
area of 102 m2, using a frequency range (band-
width) of 106 Hz.
Far from
Telescope SOLUTION
The total power received is the power/area/Hz, mul-
tiplied by the frequency range (in Hz), and the sur-
(a) (b)
face area of the telescope:
Fig 4.26. Resolution for a radio telescope. (a) The short
W/m2/Hz)(106 Hz)(102 m2)
P (10
dashed lines show what would happen if there were no
diffraction. Only radiation traveling parallel to the telescope 18
10 W.
axis would reach the focus.The solid lines show the effects of
This is 10 20 of the power of a 100 W light bulb.
diffraction. Radiation coming in at a slight angle with the tele-
scope axis can still be re¬‚ected on to the focus.This means Note that the larger the dish, the larger the total
that when the telescope is pointed in one direction, it is sen-
power detected.
sitive to radiation from neighboring directions.This is shown
in (b), as the telescope is sensitive to radiation coming from We have already seen that for making maps of
within a cone of angle approximately /D (in radians). extended sources, larger dishes are important
because they provide us with better angular reso-
lution. The weakness of radio sources gives us
a focal point. (They are like optical reflectors.) The
another reason for building large telescopes. A
long wavelength becomes important in this
larger telescope intercepts more of the radiation,
process. We have already seen that the resolution
and allows us to detect weaker sources. A few
of a telescope depends on the size of the tele-
large telescopes are shown in Fig. 4.27. The largest
scope, relative to the wavelength (Fig. 4.26). (In
single dish is at the National Atmospheric and
the radio part of the spectrum, atmospheric see-
Ionospheric Center in Aricebo, Puerto Rico (Fig.
ing is not a problem.) Since the wavelengths are
4.27a). The dish is made of a mesh surface that is
large, to achieve good resolution you need a large
set in a natural bowl. The holes in the surface are
collector. However, that surface doesn™t have to
large enough that we can only use this dish for
be perfect. It can have imperfections as long as
long wavelength observations. Also, the dish can-
they are smaller than approximately /20. For
not be steered in any direction; it looks straight
example, at a wavelength of 20 cm, 1 cm diame-
up. However, by moving the detectors around you
ter holes have no effect on the performance of
can actually view a reasonable amount of sky. For
the telescope. We are hindered by the fact that it
many years, the largest fully steerable antenna
is hard to make large telescopes with very accu-
was at the Max Planck Institüt für Radioastromie in
rate surfaces. Most large telescopes are made up



Fig 4.27. Large radio telescopes. (a) The 300 m (1000 ft )
dish in Aricebo, Puerto Rico.The dish always points straight
up, but moving the receiver to different off axis positions
allows looking away from overhead. (b) The 100 m diameter
telescope of the Max Planck Institüt für Radioastronomie,
Effelsberg, Germany. It operates in azimuth and elevation.
Azimuth is controlled by moving the whole structure on the
circular track.As the telescope changes its elevation angle it
deforms under gravity. However, it is designed to deform from
Effelsberg, Germany (Fig. 4.27b). That telescope is one paraboloid into another, so only its focal length changes.
100 m in diameter. The inner 80 m has a surface (c), (d) The 100 m Byrd Telescope at the NRAO in Green
made of solid panels so it can be used at wave- Bank,WV. The offset arm to support receivers results in no
lengths of a few centimeters. The outer 20 m is a blockage of the dish.This optimizes sensitivity and imaging
quality.The telescope surface (c) and back-up structure (d) are
mesh and is used to make the dish larger at
shown. [(a) The Aricebo Observatory is part of the National
longer wavelengths where diffraction is worse.
Astronomy and Ionosphere Center operated by Cornell
The newest large telescope is the Byrd telescope,
University under a cooperative agreement with the National
just being completed at the NRAO in Green Bank.
Science Foundation; (b) MPIFR; (c),(d) NRAO/AUI/NSF]
It is a fully steerable 100 m telescope (Fig. 4.27c). It
is designed so that more of the collecting area is
used than in the German telescope. This is done structure. All of the surface is accurate enough for
in part by reducing the blockage by the support observations at wavelengths as short as 7 mm.

The detection of the radio waves takes place in of the signal, and doing this for different delay
a radio receiver. In general, the size of the receiver times. We can think of this as the digital analog
limits us to only one receiver operating on a tele- of the Michelson interferometer spectrometer
scope at a given time. This is equivalent to doing discussed above. It therefore produces a Fourier
optical observations with only one detector in transform of the spectrum. As with the
your CCD. At any given time, the telescope is Michelson interferometer, it is limited by its abil-
receiving radiation from a piece of sky determined ity to measure this Fourier transform at only a
by the diffraction pattern of the antenna. If we finite range of time delays, with a step size lim-
want to build up a radio image of part of the sky, ited by how fast we can run the computer. The
we must point the telescope at each position and computer speed determines the total bandwidth
take a separate observation. Recently, improve- of the spectrometer, and the largest delay time
ments in receiver technology have allowed limited determines the frequency resolution. By chang-
multireceiver systems. ing one or the other, we can adjust the band-
The receivers in radio astronomy are similar in width or the frequency resolution. That is why we
concept to home radios. Like your home radio, the say this is a very flexible system.
incoming signal is first mixed with a signal from With either technology it is relatively easy to
a reference oscillator, and the resulting lower fre- make high resolution spectral observations, with
quency beat note is then amplified. We change up to a few thousand frequency channels
the frequency we are observing (like changing observed simultaneously. So, compared with opti-
radio stations) by changing the frequency of the cal observations, in radio observations we have to
reference oscillator. However, the signals from work harder to build up an image, but it is easier
astronomical sources are so weak by the time they to make a spectrum at each position in our image.
reach us that receivers for radio astronomy must One of the most important advances in radio
be much more sensitive than your home radio. astronomy in the last three decades of the 20th
Sometimes the receivers are cooled to a few century has been the development of the millime-
degrees above absolute zero to minimize sources ter (or shortest radio wave) part of the spectrum.
of background instrumental noise. Unlike bolome- As we will see throughout this book there are cer-
ters, they do not simply detect all of the energy tain observations which are only possible at mil-
that hits them; they are also capable of preserving limeter wavelengths. There also are some inherent
spectral information. benefits in working at millimeter wavelengths. If
Just as with optical observations, in radio we can observe at 1 mm, for example, we can
astronomy we can make continuum and spectral achieve the same angular resolution as at 10 cm,
line observations. Continuum studies are like opti- with a 100 times smaller dish! Of course, the dish
cal photometry. We tune our receivers to receive must have a surface that is 100 times more accu-
radiation over a wide range of frequencies, and we rate, meaning that it is hard to make a very large
measure the total amount of power received. From dish. This has restricted the size of millimeter tel-
this information, we obtain the general shape of escopes to a few tens of meters in diameter, pro-
the continuous spectrum (intensity vs. frequency). viding resolutions of 10 to 20 arc sec at best.
In spectral line observations the radiation is At millimeter wavelengths the atmosphere
detected in small frequency intervals, so the blocks some of the incoming radiation (being
shapes of spectral lines can be determined. The somewhere between the totally clear radio and
spectrometers for radio observations have tradi- totally blocked infrared). This means that it is
tionally been large numbers of electronic filters, useful to put millimeter telescopes at high alti-
tuned to pass narrow frequency ranges. More tudes and dry sites (just as with optical or
recently, the ability of very fast computer chips infrared telescopes). One of the first (and until its
has allowed for very flexible digital spectrome- closure, in 2000, one of the most heavily used)
millimeter telescopes was the 12 m telescope of the
ters. They measure the auto-correlation function
of the incoming signal, which is the result com- NRAO, located on Kitt Peak, Arizona, just below
paring the signal with a slightly delayed version the site of the optical observatory (Fig. 4.28a). ESO



(b) (d)
Fig 4.28. Millimeter telescopes. (a) The NRAO 12 m telescope on Kitt Peak, Arizona. (b) The Swedish“ESO Submillimetre
Telescope (SEST) on La Silla, Chile. (c) The 30 m telescope, Spain. (d) The Nobeyama 45 m telescope. [(a) Jeffrey Mangum,
NRAO/AUI/NSF; (b) ESO; (c) IRAM; (d) Nobeyama Radio Observatory/National Astronomical Observatory of Japan]

has operated the 15 m Swedish“ESO Submillimetre using combinations of telescopes, called interfer-
Telescope (SEST) at their optical site on La Silla, ometers (Fig. 4.29). Interferometers utilize the
Chile (Fig. 4.28b). The largest millimeter tele- information contained in the phase difference
scopes are the 30 m telescope operated by French between the signals arriving at different tele-
and German institutes, and located in Spain scopes from the same radio source. Any pair of
(Fig. 4.28c) and the Nobeyama 45 m telescope in telescopes provides information on an angular
Japan (Fig. 4.28d). scale approximately equal (in radians) to the
The problem of poor angular resolution for wavelength, divided by the separation between
radio observations has been solved, in part, by the two telescopes in a direction parallel to a line

ence between signals from objects at the center

of each field of view. But objects off the field cen-



ter will have varying phase differences as we


track the source across the sky. To extract a map

of our field of view, we first multiply the two sig-

nals together (actually the delayed signal from

the first telescope times the complex conjugate
of the signal from the second telescope). This
product is called the visibility. When you work out
the details, the visibility turns out to be the
Fourier transform of the two-dimensional inten-
sity distribution on the sky, I(x, y). The visibility

is a function of two variables (u, v), where u
(d/ )cos 0 and v is defined for the corresponding
θ0 angle in the perpendicular direction.
d So, we measure the (2D) visibility at as many
(u, v) points as possible, and then calculate I(x, y)
Fig 4.29. Radio interferometer. Here only two telescopes
by taking the 2D Fourier transform. Obviously,
are shown, but an interferometer with any number of tele-
the more (u, v) points we can measure, the better
scopes can be treated as a number of pairs of telescopes.
we can estimate the visibility and the better we
The separation between the telescopes produces a phase
can estimate its Fourier transform. This is similar
delay, which depends on the separation, d, and the position
to the Michelson interferometer, discussed ear-
of the source.The phase difference can be detected, provid-
lier in this chapter, where the more mirror posi-
ing information about source structures whose angular size
is approximately /d (in radians). By using different telescope tions at which we could measure the interference
spacings and the Earth™s rotation, information about struc- pattern, the more accurately we could compute
tures on different angular sizes can be accumulated and
the spectrum, which is the Fourier transform.
eventually reconstructed into a map of the source.
How do we measure many (u, v) points? For
any pair of telescopes, we let the Earth™s rotation
change the elevation angle of the source, and also
connecting the two telescopes. To obtain infor-
the orientation on the sky, changing how much
mation on different angular scales, it is necessary
of u and how much of v we are changing. So, if we
to have pairs of telescopes with different spac-
do a series of observations, of say, 5 minutes each,
ings. In addition, different orientations are
and track a source for 8 hours, we can take many
needed. For this reason, interferometers gener-
measurements. It also helps to have many tele-
ally have a number of telescopes. The Earth™s rota-
scopes. For N telescopes there are N(N 1)/2, inde-
tion also helps change the orientation of any pair
pendent pairs of telescopes, so, for large N, the
of telescopes, as viewed from the direction of the
number of pairs goes up roughly as N2. To make
source. Unlike single dish observations, you don™t
optimum use of these pairs, we don™t simply have
have to point the telescope at different parts of
equally spaced telscopes, since every pair of spac-
the source to make a map.
ing d will give redundant information. It is also
To see some of the limitations of using inter-
useful to not have all the telescopes in a
ferometers to make images, we look a little more
line,which would just give a lot of values of u or
at how they work. We again look at any pair of
v, but not both. In general the shortest spacings
telescopes, as indicated in Fig. 4.29. Before com-
give information on the large angular scales on
bining the signals from the two telescopes, we
the sky, and the longest spacing provide informa-
delay the signal from the nearer telescope by the
tion on the smallest angular scales.
extra time it takes the waves to reach the second
The most useful interferometer over the past
telescope. That delay will change as we point the
several years has been the Very Large Array or VLA,
telescope pair at different angles above the hori-
near Socorro, New Mexico, operated by the NRAO
zon. This allows us to zero out the phase differ-


Fig 4.30. Views of the Very Large Array (VLA), on the Plains of St Agustine (at an altitude of about 2.3 km) southwest of
Socorro, New Mexico.There are 27 telescopes, each 25 m in diameter. At any instant there are 351 pairs of telescopes.
Depending on the project, the spacings can be adjusted by moving the telescopes along railroad tracks. Moving all of the tele-
scopes takes a few days and is done four times a year.The VLA is operated by the National Radio Astronomy Observatory.
(a) The whole array. (b) The central section, showing a better view of the telescopes and the railroad tracks. (c) The transporter
used to move telescopes on the tracks. [NRAO/AUI/NSF]

(Fig. 4.30). The VLA has 27 telescopes, 351 pairs a few hours. (The amount of data taken is so large
(each 25 m in diameter), arranged in a ˜Y™ config- that it takes the computers much longer to
uration, to allow a wide range of both (u, v) values. process the data than it does to observe.)
Each arm of the Y is 21 km long. The telescopes For observations requiring the best possible
are placed alongside railroad tracks, so that the resolution, telescopes on opposite sides of the
Earth are used. This is called very long baseline
telescope spacings can be changed, depending on
interferometry (VLBI). VLBI observations have pro-
the resolution needed for a particular project.
vided angular resolutions of 10 4 arc seconds! In
These changes can take up to two weeks and are
only done a few times a year. The shortest wave- regular interferometry, the signals from the vari-
length at which the VLA operates is 7 mm. It can ous telescopes are combined in real time, as the
be used for both continuum and spectral line data comes in. In VLBI, signals at each telescope
observations. It has proved to be a powerful tool, are recorded along with a time signal from a very
providing images of radio sources, with many accurate atomic clock. Later, the records are
observing sessions ranging from a few minutes to brought together, and the time signals are used

has been chosen. It is expected that operation
will start in 2010.

4.9 High energy astronomy

X-ray astronomy is one of the youngest fields in
observational astronomy. Since X-rays do not
penetrate the Earth™s atmosphere, the history of
X-ray astronomy is the history of high altitude
(balloon) and space astronomy. Early X-ray obser-
vations were done with sounding rockets (which
provided very brief flights with only a few min-
utes of data taking) and high altitude balloons. Of
course, the balloons still do not rise above all the
atmosphere, and, in the X-ray part of the spec-
trum, even the little bit that is left matters.
Fig 4.31. Arrangement of telescopes for the Very Long
One problem in observing X-rays is that it is
Baseline Array (VLBA), operated by the NRAO.There are
very hard to make a mirror that works for these
ten telescopes, each 25 m in diameter. It operates down to a
short wavelengths, less than 1 nm. That is
wavelength of 1 cm. [NRAO/AUI/NSF]
because the typical spacing between atoms in a
solid is about 0.1 nm. So the incoming radiation
sees a rough surface, with reflections off each
atom producing a scattering in some essentially
to coordinate the records from different tele-
random direction. There is one possibility. If we
scopes, and the interferometry is then done by
arrange for the X-rays to come in at a very shal-
computer. To provide a dedicated group of tele-
low angle, only a degree or two, with the surface,
scopes for VLBI, the NRAO has recently built the
Very Long Baseline Array (VLBA), which extends over the atoms appear to be closer together, and we
achieve normal reflection. This is called grazing
much of North America (Fig. 4.31).
incidence. Of course, being constrained to only graz-
The success of interferometry and the impor-
ing angles makes it difficult to design a telescope
tance of millimeter observations have led
that will collect and focus a reasonable amount of
astronomers to begin working with millimeter
radiation. A diagram of the imaging system in one
interferometers. Millimeter waves provide many
X-ray satellite is shown in Fig. 4.33(b).
technical challenges for interferometry. For
X-ray satellites are able to provide both con-
example, the effects of the Earth™s atmosphere
tinuous and spectral line observations. Originally,
on the millimeter signals are important.
the spectral information came from detectors
Following the demonstration of successful mil-
similar to those used by high energy physicists,
limeter interferometers operated by Caltech and
called proportional counters, which register the
Berkeley, the NRAO has started development of
the Atacama Large Millimeter Array (ALMA), shown energy of the photons as they hit. Better spectral
resolution was obtained by using a type of grating
in Fig. 4.32. The final details are still being
called a Bragg crystal, in which the ˜slits™ are the
worked out, but the array, which is being built in
individual atoms in a solid. More recently, X-ray
collaboration with ESO and the Chilean govern-
astronomers have been able to use solid state
ment, will have approximately 75 telescopes,
detectors that give good spectral resolution.
each approximately 12 m in diameter. They will
More recently a number of satellites have
work down to a wavelength of 0.8 mm. In order
opened our eyes to the X-ray sky. One of the satel-
to get around atmospheric problems, a very high
lites is shown in Fig. 4.33(a). A number of the
(5000 m) dry site in the Atacama Desert in Chile


tions was Uhuru (launched in 1970). It was also the
first to survey the whole sky, and, compared to
previous missions, had very sensitive detectors. It
found 339 objects, showing astronomers that
many different types of objects give off strong X-
ray emission. Following Uhuru, which was a rela-
tively small satellite, NASA launched a number of
larger satellites in the High Energy Astronomy
Observatory (HEAO) program. The second in the
HEAO series was the Einstein Observatory, which
(b) was the first to utilize the grazing incidence
imaging, and so produced the first real X-ray
images. The Einstein images had a profound
impact on our thinking about many types of
astronomical objects. We went from being able
to probe small sections of objects to forming
whole images. In many ways it was like having a
blindfold removed. The next major jump in sen-
sitivity and angular resolution was the Roentgen
Satellite (ROSAT), launched in 1990 (Fig. 4.33a).
Chandra was launched in 1999, and provides
sub-arcsecond imaging (Fig. 4.33c) and grating
spectrometry, so it does high quality imaging
spectroscopy (Fig. 4.33d).
All of the telescopes that we have discussed so
Fig 4.32. The Atacama Large Millimeter Array (ALMA).
far have been for electromagnetic radiation. High
(a) Views of the site. (b) Artist™s conception of the
arrangement of telescopes. (c) Artist™s conception of the energy phenomena also make their presence
antenna appearance. [NRAO/AUI/NSF and ESO] known in other ways. One way is by the emitting
beams of cosmic rays, charged particles, often
earliest satellites incorporated X-ray detectors, with very high energies. They also give off neutri-
which shared space with other experiments. The nos, subatomic particles that are very difficult to
first satellite devoted entirely to X-ray observa- detect. They also give off gravitational radiation,

distortions in the fabric of spacetime, which Paraboloid Hyperboloid
Surfaces Surfaces
again are very difficult to detect. We will talk
about detecting each of these in the chapters that X-rays

discuss their astrophysical origin.
Focal Point




Fig 4.33. X-ray satellites.
(a) ROSAT. (b) Imaging system in
the Chandra X-ray satellite, utiliz-
ing grazing incidence. (c) Chandra
image of the Crab Nebula showing
the great detail achievable in cur-
rent X-ray systems. (d) Chandra
test spectrum, showing the good
spectral resolution. [NASA]


Chapter summary
In this chapter we saw how different types of tel- the radio), and in other cases is limited by atmos-
escopes are used to collect data across the elec- pheric seeing (especially in the visible).
tromagnetic spectrum. We saw the differences We saw the various techniques for extracting
and similarities among the techniques used in information from the radiation collected by our
different parts of the spectrum. Much of the telescopes. Improving detector efficiency and
progress in astronomy in the past few decades panoramic ability has been important in all parts
has come from our ability to make high quality of the spectrum.
observations in parts of the spectrum other than We saw the importance of site selection for an
the visible. observatory. As an ultimate site, we saw the
We looked at the important features of any advantages of telescopes in space. In space, we
telescope, the collecting area and the angular res- can observe at wavelengths where the radiation
olution. The collecting area determines how sen- does not penetrate the Earth™s atmosphere. Even
sitive the telescope is to faint objects. The angular in the visible, we can achieve improved sensitivity
resolution is limited by diffraction (especially in and angular resolution.


think this works? What are the possible draw-
4.1. Describe the factors that limit the angular
backs to this?
resolution of an optical telescope. Include
4.7. What are the advantages of CCDs over photo-
estimates of the size of each effect.
graphic emulsions and photomultipliers?
4.2. What do we mean when we say that the
4.8. Compare image formation (similarities and
main reason for building large ground-
differences) in the eye and in a camera.
based optical telescopes is light-gathering
4.9. Why is chromatic aberration a problem even
for black and white photographs?
*4.3. Explain how improving the seeing at a site
4.10. If there is no angular magnification in a sim-
might allow you to detect fainter stars. (Hint:
ple camera, how can using a longer focal
Think of what happens to the photons on
length lens give a larger image?
your detector (film or CCD) when the image
4.11. A higher quality (more expensive) camera
is smeared.)
lens generally has smaller f-stops than an
*4.4. Suppose you are observing two stars that are
inferior lens. Why is this?
2 arc sec apart. Draw a diagram illustrating
4.12. Why do some people need to wear eyeglasses
what you observe under conditions of 4, 2
while driving at night but not during the
and 1 arc sec seeing. Your diagram should be
day? (What is it about the lower light level
a series of graphs showing intensity as a func-
that degrades images?)
tion of position on a detector.
4.13. Compare the advantages and disadvantages
4.5. Suppose two stars are 5 arc sec apart on the
of reflecting and refracting telescopes.
sky. We clearly cannot resolve them with our
4.14. Compare the advantages and disadvantages
eyes, but the angular resolution of even a
of various focal arrangements in reflecting
modest sized telescope is sufficient to resolve
them. However, the light from the telescope
4.15. What does it mean to focus the eye or a
must still pass through the narrow pupil of
camera ˜at infinity™?
the eye. Why doesn™t the diffraction of the
4.16. If you want to photograph a planet you use
light entering the eye smear the images too
your long focal length telescope; if you want
much for us to resolve the two stars?
to do photometry on a faint star, you use a
4.6. ˜Faster™ photographic emulsions can be made
short focal length telescope. Explain.
by making the grains larger. Why do you

4.26. Explain the similarities between trying to dis-
4.17. For many observations (both imaging and
play astronomical images and displaying
spectroscopy) it is becoming important to
things such as topographical maps.
read the data into a computer. Briefly discuss
4.27. Explain why simply using color film in your
the techniques for doing this that we have
camera does not give you a true color astro-
mentioned in this chapter.
nomical photograph.
4.18. What are the important considerations in
4.28. In what ways are radio observations similar
choosing an observatory site?
to and different from (a) optical and
4.19. What are currently the best methods of
(b) infrared observations?
reducing seeing effects at a given site?
4.29. Suppose we want to use a radio telescope
4.20. What advantages does HST have over
with a transmitter rather than a receiver of
ground-based telescopes for optical
radio waves. Draw a diagram (similar to Fig.
4.26) showing how the transmitted radiation
4.21. What are the similarities and differences
would be spread out on the sky.
between ultraviolet and optical
4.30. Why is possible to observe with the Aricebo
dish even though the surface has holes in it?
4.22. What are the similarities and differences
4.31. Why is it possible to do radio observations
between infrared and optical observations?
during the day, but not optical observations?
4.23. For infrared observations, we must still live
4.32. What are the advantages of observing in the
with the fact that parts of a telescope
millimeter part of the spectrum? What are
radiate like blackbodies at about 300 K.
the additional difficulties?
Why isn™t this a problem for optical
4.33. How would an image made by the VLA compare
with one made with a single telescope as large
4.24. How does a bolometer work? How would you
as the VLA (assuming you could build one)?
use a bolometer to measure the power
4.34. How does VLBI differ from normal
received in a small wavelength range, for
example between 10 and 11 m?
4.35. What are the difficulties in making a mirror
4.25. Why can™t balloons get above all of the
to work for X-rays?


. 4
( 28)