Notes. The errors on the fluxes are the (statistical combined with systematic) 1σ errors as obtained using the uvotsource tool. The luminosities are calculated using the distance of 4.69 ± 0.04 kpc and the effective filter bandwidth of uvm2 of 533 Å, and the errors are derived by propagating the errors on the flux and the distance. The XRT upper limits were determined using the HEASARC tool Webpimms and appropriate estimates of N_H column densities and the power law index (see Sect. 3.1 for details). For SW4, no XRT upper limit could be determined due to contamination at the source location from a very bright nearby X-ray source (namely X9; see Sect. 4). The Chandra X-ray luminosities are taken from Heinke et al. (2005) for the likely counterparts to SW1-4, namely W25, W56, W324, and W30, respectively. We note that they correspond to a slightly different energy range (0.5–6 keV vs 0.5–10 keV, which we have used) and were converted to the 4.69 kpc distance used in our paper (Heinke et al. 2005 used a distance of 4.85 kpc).