Annu. Rev. Astron. Astrophys. 1992. 30:
653-703 Copyright © 1993 by Annual Reviews. All rights reserved |
3.3 Observations on angular scales 1-20°
There have been many searches for anisotropy on angular scales 1-20° since 1983. In keeping with the discussion of previous sections, we discuss these below in order of increasing angular size. It should be borne in mind that over this range of angular scales many models of galaxy formation predict angular spectra in the temperature fluctuations which are not well approximated by a Gaussian distribution. In particular cold dark matter models (e.g. Bond et al 1991) predict a strongly non-Gaussian form for W(k) over this range of angular scales. Thus, over this range of scales the observations must be compared to specific models in order to estimate the expected variance in sky temperature as given by Equation 3. In the absence of any accepted model for galaxy formation it has become common practice amongst observers to assume, for want of a better alternative, that the form of W(k) is Gaussian in order to give some idea of the sensitivity levels of their observations. We follow this practice below, but caution the reader that limits and expected values of the sky variance derived by this method cannot readily be compared to theoretically predicted levels of the sky fluctuations.
3.3.1
The observations were carried out in January and February 1987.
Timbie and Wilkinson observed fields straddling the North Celestial
Pole at declination =
87°.5. The interferometer was
switched in azimuth between the two positions 2°.5 from the pole
every 10 seconds. The output was therefore equivalent to a double
switching experiment, with the first level of switching being done by
the interferometer at a frequency comparable to the reciprocal of the
bandwidth. This double differencing eliminates long term drifts in
the signal level caused by instrumental drifts in gain and offset and
atmospheric changes.
Observations were made almost continuously over a 15-day period. The
data were binned in 12 two-hour bins in sidereal time. In two hours a
field 2°.5 away from the pole drifts about 1/3 of a
beamwidth of the main lobe in elevation; so that the sky was slightly
oversampled by this procedure. The symmetry of the observing
technique ensured that any true sky fluctuations would average to zero
when all the data were added together because any patch of sky was
observed once with positive signal and twelve hours later with
negative signal. In fact a mean value of 190 ± 50 µK was
recorded, and was ascribed to non-uniform ground spillover. This has
been subtracted from the data. The resulting measurements of this
experiment are shown in Figure 3a.
For a fit to a constant T, 2 = 1.5,
which implies
that some ``signal'' has been detected. Timbie and Wilkinson do not
interpret this result as a real detection of anisotropy, but they use
it to place more conservative upper limits on the anisotropy using the
likelihood ratio test. Their 95% confidence upper limits are given
in Table 4. The power of the test is 55%.
A second set of observations has been carried out using a more
sensitive receiver. The results will be reported in Jarosik et al
(1992, in preparation).