The sources observed are those listed in the NRAO catalogue for which flux density measurements at 750 and 1400 MHz are available (PAULINY-TOTH et al. 1966). As the observations for that catalogue were based on the source positions of the 3C and 3 CR catalogues, the present data for 2695 MHz cannot be considered to be a complete or homogeneous sample of sources.
The observations were carried out during Feb.-March 1967 and Aug. 1967-May 1968. Each source was measured during a drift scan at constant declination in the meridian or close to the meridian and the measurements were preceded or followed by the internal calibration from the gas discharge tube. In a drift scan the mean of the background radiation before and after the source was taken, so that the confusion did not set the limit of measuring accuracy. On the analogue record interference pulses and atmospheric disturbances could readily be recognized.
There existed no difficulty in pointing accuracy. The Positions of the radio sources in the NRAO catalogue have uncertainties which are one half of the position uncertainty of the Stockert telescope. To eliminate any Possible instrumental effects depending on declination, a number of calibration sources with different declinations were chosen, the list of which is given in Table 1.
Table 1. Pointing Calibration Sources --------------------------------------------- Source RA1950 Err DEL1950.0 Err --------------------------------------------- NRAO 319 09h15m40.8s 1.5s -11d53'00" 20" NRAO 236 06 24 43.2 1.0 -05 51 06 20 NRAO 400 12 26 31.1 1.5 +02 19 38 20 NRAO 152 03 56 10.0 1.0 +10 17 17 20 NRAO 229 06 06 40.4 1.5 +20 29 22 20 NRAO 187 04 33 55.6 1.5 +29 34 13 20 NRAO 171 04 15 02.1 1.0 +37 55 02 20 NRAO 285 08 09 59.5 1.5 +48 21 47 20 NRAO 437 14 09 34.1 1.5 +52 26 00 20 ---------------------------------------------For these sources the positions were determined as accurately as possible with the Stockert telescope and the results were compared with those given in Table 1. No systematic difference with declination could be seen. Hence all sources were measured with drift curves at the declinations given in the NRAO catalogue and no further determinations of positions for individual sources were considered necessary.
Table 2. Flux-density Calibration Sources ------------------------------- Source Flux Density (FU) ------------------------------- NRAO 187 27.24 NRAO 319 23.33 NRAO 57 7.94 NRAO 79 9.18 NRAO 152 7.05 NRAO 171 10.52 NRAO 221 13.18 NRAO 285 7.70 NRAO 395 13.20 NRAO 400 38.40 NRAO 437 11.75 NRAO 518 22.20 NRAO 524 35.95 NRAO 658 6.48 -------------------------------Flux density values for the first two sources in Table 2 were taken from KELLERMANN et al. (1968). These two sources were chosen as the main standards because they are relatively intense, less than 1% linearly polarized, used by most observers at other frequencies, and are not observed to be variable. The other standard sources were measured relative to these two sources. It should be stressed that all measurements are with the E-plane of the feed horn oriented east-west. Care should be taken when adopting the same flux values for future measurements; NRAO 524 has been corrected for source size and NRAO 79, 285, and 400 are variable or suspected to be variable. During the observing time, however, no significant variation could be seen. During each night of observation at least one of the sources in Table 2 was measured and the data of that night were corrected according to the deviation from the calibration source value listed in Table 2.
The flux density error includes the standard error, a 2% error of the total flux density plus an estimated general error to account for the low number of repeated measurements. As mentioned above no error to account for the confusion limit was taken. As each source with a flux density greater than 0.4 FU was measured on two or three different days, and each source with a flux density less than 0.4 FU was measured up to five times, the flux error for sources between 0.2 and 1.0 FU is about the same. Corrections for source size have been made if necessary. To calculate the spectral index the 1400 MHz fluxes were taken from the NRAO catalogue. Correction factors to the original values of this catalogue have been applied (KELLERMANN, PAULINY-TOTH, WILLIAMS, CASWELL 1968).
Fig. 2. Histogram of the spectral indices for quasistellar sources. Objects with a flux density S (2695 MHz) > 1 FU are shaded. Upper diagram for the frequency range 750-1400 MHz, lower diagram for 1400-2695 MHz. a) Total number of sources, b) Sources with apparently straight spectra, c) Sources of the class G_. d) Sources of the class C+.
Fig. 3. Histogram of the spectral indices for radio galaxies.
Fig. 4. Histogram of the spectral indices for unidentified radio sources.
The number of sources, which show curved spectra beyond calibration and measurement errors, is still too low for an analysis of their spectral distribution. A simple classification of the spectra is given in Tables 4 and 5.
Table 4. Number of sources with a simple or curved spectrum, values for sources with S_nu(2695 MHz) > 1 FU in parentheses ---------------------------------------------------- simple C_ C+ ---------------------------------------------------- Quasar 62 (48) 15 ( 9) 7 (7) Radiogalaxy 111 (89) 29 (25) 10 (10) Unidentified Source 280 (74) 55 (20) 31 (12) ---------------------------------------------------- Table 5. Number of different types of galaxies with a simple or curved spectrum. ---------------------------------------------------- Galaxy of E D N DB S IRR SEY unknown type ---------------------------------------------------- simple 48 23 24 9 4 2 1 C_ 11 6 4 3 2 1 1 1 C+ 1 7 1 ----------------------------------------------------It is apparent that quasars and E-galaxies are the, main contributors to sources with flat spectra. As such sources were more likely to be missed by the low frequency surveys, a search for new quasars should be made at high frequencies.
The homogeneous sample within the NRAO sources are the 3C objects, for which the detection limit was 9 FU at 178 MHz (BENNETT 1962). Extrapolating with a spectral index of -0.8 to 2695 MHz yields 1 FU. So all sources with S (2695 MHz) > 1 FU are essentially the homogeneous sample of Table 3.
In Table 4 we see for S (2695 MHz) > 1 FU a difference between quasars and galaxies in the number of sources for the classes C_ and C+. It is tempting to use this difference to predict, whether there are more galaxies or quasars among the unidentified sources. Taking the ratio for quasars and galaxies as given in Table 4 for S (2695 MHz) > 1 FU we calculate for 74 unidentified sources only about 14 for the class C_ and 7 for the class C+. The actually observed number of unidentified sources with curved spectra, however, is almost twice as high in both classes (Table 4). This may indicate that besides quasars and galaxies another contributor is present, perhaps simply galactic objects. Histograms of the spectral index for the various types of radio sources are given in Fig. 2-4. The distribution for the 750/1400 MHz data is shown as well as the distribution for the 1400/2695 MHz data. The error of the spectral index of a source depends upon its flux density, especially the 2695 MHz one. To give an idea of distribution for the strong sources; the histogram for sources with S(2695MHz) >= 1 FU is shaded. The contribution from sources that seem to obey a simple power law and from sources with Deepening and flattening spectra are shown separately. From the theory of synchrotron emission we expect a greater diversity of spectral indices at higher frequencies. For sources with curved spectra this has already been shown by our measurements, for the rest a greater dispersion at higher frequencies should be seen. Before the histograms of Fig. 2-4b may be compared, we have to correct for the different measurement errors. If am is the dispersion of the spectral index and An the mean error of the spectral index (both calculated from Table 3), the corrected dispersion an is calculated from sigma_n = sqrt(sigma_m^2 - Delta n^2) The implication for such a simple correction is that the distribution functions of both the spectral index and its error are Gaussian, an assumption which is a fair approximation. Then we obtain Table 6.
Table 6. Corrected dispersion of the spectral index for sources with S (2695 MHz) >= 1 FU and no curved spectrum. sigma_n(750/1400) sigma_n(1400/2695) -------------------------------------------------------- Quasars 0.29 0.25 Galaxies 0.16 0.14 Unidentified Sources 0.19 0.24 --------------------------------------------------------Quasars and galaxies do not show the predicted increase in dispersion at the higher frequencies, and for the unidentified sources the dif ference is not significant. - From Table 4 we attempted to predict the nature of the unidentified sources with curved spectra and obtained negative results. Trying again for sources with apparently straight spectra from Table 6 we cannot draw a conclusion either, as the dispersion of the unidentified sources for the low frequency range indicates galaxies and for the high frequency range indicates quasars. Optical identifications are generally available only for the strong sources, which were originally in the 3 C catalogues. The search for differences between the radio properties of radio galaxies and quasars was not fruitful. No difference in their spectral distribution can be seen, except that the dispersion of the quasars is greater (Table 6). This is consistent with the Green Bank results, when they are restricted to 3 C objects; when quasars of the Parkes catalogues are included, however, the nature of the distribution changes and the median moves towards a greater value (KELLERMANN et al. 1968). The basic survey at Parkes was made at 408 MHz whereas the 3C surveys were made at 159 and 178 MHz. As no principal difference for the northern and southern hemisphere can be expected, the discrepancy between the two surveys stresses the need for more surveys at different frequencies. Weak galaxies and quasars tend to have steep spectra (Fig. 2-3). This is a selection effect: 1 FU corresponds to the detection limit of the 3 CR survey. Hence, sources that were just above the detection limit of the 3 CR survey and that have steep spectra appear in the present sample as weak sources, whereas sources that were just below the detection limit of the 3 CR survey and that have flat spectra, and are also above the detection limit of the 2695 MHz receiver, are excluded.
The distribution of the spectral index for the unidentified sources shows one marked difference from either the quasars or the galaxies. In the lower frequency range the strong sources tend towards flat spectra and in the higher frequency range towards steep spectra; for weak sources the opposite is true (Fig. 4). It is not certain whether this is due to the inhomogenous sample of sources. But one effect can contribute to the distribution observed: On the average strong sources are near and old, and weak sources are further away and younger. From the theory of synchrotron emission we know that the spectrum of a source steepens with age, if there is no injection of new relativistic electrons, and that the process is moving from high frequencies to lower frequencies. Thus strong radio sources on the average can have steeper spectra at higher frequencies. As for the distribution of the weak sources, they were primarily found by chance during the 750 and 1400 MHz measurements at Green Bank. Remembering that the Parkes survey for quasars at 408 MHz indicates a shift of the median towards a greater value, we can expect a similar effect for the random Green Bank sample. Recent limited surveys at 5000 MHz also show the tendency of weak sources towards flat spectra (BLUM, DAVIS 1968; KELLERMANN, PAULINY-TOTH, DAVIS 1968), and KOMESAROFF et al. (1968) concluded from a comparison of optical and radio data for clusters of galaxies that the weaker radio emitters have flatter spectra.