6 Factors affecting the accuracy

6.1 Zero-level offsets

The offsets of the T-T plots between 408 MHz and the final 22 MHz map are in the range $\pm$ 10 kK at 22 MHz. The mean of 11 offset values at intervals of 10$^\circ$ in declination from +79$^\circ$  to -21$^\circ$  is -5.1 kK with an rms deviation of 11.1 kK. This mean offset is of the same order as that derived for the T-T plot of zenith data (see Fig. 1). From these data alone, however, it is not possible to decide the extent of the true offsets in either the 22 MHz or the 408 MHz data. The 408 MHz data is suspected of having offsets as large as 2 K (Reich & Reich 1988) and may also have a baselevel correction of 5 K. If the average spectral index between 22 and 408 MHz is 2.50, these values correspond to 2.8 kK and 7.2 kK at 22 MHz.

6.2 Effects of antenna properties on the data

In this section we examine the possible effects on our data of our imperfect knowledge of the properties of the antennas. The antennas used at 22 MHz and 408 MHz are of different types, and need to be discussed separately.

If emission received in the sidelobes made a significant contribution to antenna temperature, then errors would result. These would be especially significant in the map of spectral index, because the sidelobe responses of the two very different antennas would receive emission from different parts of the sky. The greatest effect would occur in measurements of those parts of the sky where the brightness temperature is the lowest (around 9 hours, 30$^\circ$) with emission from the bright Galactic plane being received in the far sidelobes of the antennas. As an example of such effects, Landecker & Wielebinski (1970), using the Parkes 64-m telescope at 150 MHz, found that about one third of the antenna temperature at the sky minimum arose from sidelobe contributions.

The response of the 22 MHz antenna is, in principle, completely calculable from the geometry of the array, and from the phasing and grading applied to the array elements. The angular size of the main beam proved to be very close to the calculated value. The net sidelobe solid angle should be zero, implying a beam efficiency of 1.0. Sidelobe responses, apart from their effects near strong sources, should not affect the measurement of the broad structure which is the focus of this paper. Measurements of the antenna response using the bright sources Cas A and Cyg A (Costain et al. 1969) bear out this expectation. The dynamic range of these measurements is about 30 dB, determined by the ratio of the flux density of Cyg A (29 100 Jy) to the confusion limit for the telescope (30 Jy). Sidelobes above this level were confined to the NS and EW planes (strictly a small circle EW, depending on the phasing in declination) and were alternately positive and negative, as expected, and close to the predicted amplitude. Sidelobe response fell below the detection limit within 10$^\circ$ of the main beam in the EW direction and below 1% within 18$^\circ$ of the main beam in the NS plane. These measurements verify our assertion that the performance of the telescope at the zenith is well understood (Cas A and Cyg A pass within 10$^\circ$ of the zenith at DRAO).

We are confident that the response of the antenna to the extended background was as predicted near the zenith because of the linear relationship between the 22 and 408 MHz brightness temperatures with the expected spectral index at that declination (48.8$^\circ$). We know that the behaviour away from the zenith departed from the expected response for the extended emission features, but not for point sources. We tentatively attribute this to inadequately compensated mutual impedance effects between phased rows of dipoles in the array. At increasing zenith angles, these effects may have dominated the predicted response of individual dipoles above a reflecting screen.

The available measurements suggest, however, that the sidelobes of the complete telescope (as opposed to the individual groups of radiating elements) were still confined to the predicted regions, even at large zenith angles. If we assume that the sidelobe level was all positive and at the detection limit (-30 dB) in two 180$^\circ$ strips in the EW and NS directions, each equal in width to the main beam, then the beam efficiency would be 0.9, better than most reflector antennas. However, this is very much a worst-case assumption, and it is probably safe to conclude that the beam efficiency was $\geq$0.95. Furthermore, the largest sidelobes lie in the NS plane, and, when the main beam is measuring the coldest region of the sky, these sidelobes do not intersect the bright Galactic plane. We therefore feel justified in ignoring the effects of the sidelobe response of the 22 MHz telescope.

The 408 MHz data were not directly corrected for sidelobe contributions, but an indirect correction was applied (Haslam et al. 1982). The absolutely calibrated survey at 404 MHz made by Pauliny-Toth & Shakeshaft (1962) with a beam of about 7.5$^\circ$  was used to establish both the zero-level and the temperature scale of the 408 MHz survey data. Since the 404 MHz survey was corrected for sidelobe contributions, using it as a reference for the later survey roughly corrected those measurements for sidelobe contributions. The technique is valid in this case since the relationship between the measured antenna temperature and the sidelobe correction is, to a good approximation, linear. Checks of the effectiveness of this procedure were made subsequently by Lawson et al. (1987) and by Reich & Reich (1988) who convolved the 408 MHz data to the broad beams of horns and other low-sidelobe antennas used by Webster (1975) and Sironi (1974) to measure the Galactic emission at 408 MHz. In all cases the comparisons were satisfactory, indicating that sidelobe effects had been effectively removed from the 408 MHz data.