Discussion

The image with all pulsars still present (figure 5) shows that we do not even measure all of the flux from the known pulsars - let alone any extra from the unresolved pulsars. This is unexplained, but once the bright pulsars are also subtracted (figure 6), there is a similar amount of flux from the unresolved pulsars, to those visible on any one day. The most obvious cause of the loss of flux, is in self calibration - when calibrating for amplitude and phase, any object not appearing in the model will be scaled down. The possibilities for self-calibration are to use the 6km image as a model, and calibrate for phase only, or to calibrate for amplitude as well, making sure that the rms gain is not scaled to unity. A final alternative is to form a smaller image (because of computational requirements) using the 6th antenna for each day (or for all days combined), that will suffer because of poor UV coverage, and using that as a model in self-calibrating each day. This may have the benifit of including at least some more flux from the core, because of the presence of smaller baselines than those in the McConnell and Ables' (2000) 6km mosaic. The results from testing the peak and approximate integrated flux from each of these methods is in table 3, but since this work is incomplete, some

Table: Peak and integrated flux from different methods of self-calibration.
1: no calibration.
2: Selfcalibration, phase only, model: McConnell and Ables (2000).
3: Amplitude and phase, MIRIAD: OPTIONS=NOSCALE.
4: Amplitude and phase, model: smaller image containing 6th antenna

24cmCalibration and flux measurement on	23cmCalibration Type	23cmPeak flux ($\mu$Jy/beam)	23cmIntegrated flux($\mu$Jy)
all days combined	1	515	893
all days combined	2	422	622
all days combined	3	347	703
only 2000-Dec-27	3	465	525
only 2000-Dec-27	4	482	605

of the methods are from a previous non-optimised mosaic (without the pulsars subtracted), and some are from only one day (2000-Dec-27). It is clear that we would not have so much ``missing'' flux, if a self-calibration was performed on each day, using the full 6 antennae from that day, and so this will be used in the future. One problem faced is that there is a lack of good UV coverage, so this process will not necessarily be trivial. Incorrect flux measurements from Camilo et. al. (2000) cannot be ruled out either - if their fluxes are systematically higher, we can not expect to measure an extended component equal to the sum of pulsar fluxes for the 15 measured pulsars. Even though we do not see all the flux from the known pulsars, we do see some extra flux once the resolved ones are subtracted, that indicated we are seeing some of the extended component we are looking for, although it is may not be equal in intensity to the resolved pulsars listed in Camilo et. al, 2000, depending on the sucess of the self-calibration method proposed above. Some possible mechanisms for there not being the expected unresolved flux are summarised below. Equation 1 implies a power law with index -1. If this is not correct, and the exponent is actually more negative, then there will not be as much flux in the lower luminosity decade, and there will be less extended flux. This is also a more elegant solution than having a low luminosity cutoff around $L_{400} \sim 1$ mJy kpc$^2$[1]. If for some reason, the fainter pulsars are systematically more spread out than the sudden $3 r_c$ boundary for the brighter pulsars[2], and are distributed far enough that synthesised beam of $\sim 60$'' attenuates the integrated flux appreciably, then the flux due to the decade of low luminosity pulsars will be less than the decade of higher luminosity pulsars, although a physical reason for any possible difference in distribution has not been discovered. One promising sign is that it seems (within the error caused by noise in the images) that the integrated flux of the core once the pulsars visible on each day are subtracted, is half that of the total flux before subtraction. If indeed, we are seeing all the flux present, and the fluxes listed in Camilo et. al. are wrong, then there is a good chance the prediction of roughly 10 times more pulsars (whose beam is in our direction) being present in 47-Tuc than we currently resolve is correct - ie perhaps more than 40, since we see up to 4. Care needs to be taken however, since there may be other sources near the core that are not pulsars, and weren't subtracted, that may influence the fitting, particularly since the fit was not for a point source.