5.2. Photon Escape
As discussed above, while OII and H are not very sensitive to the fraction of escaped photons, UV is. The escape of UV photons can be quantified by the parameter [Rowan-Robinson & Efstathiou 1993] by assuming that
where (1 - ) represents the escaped fraction.
Thus, the reprocessed fraction can be estimated through the observed 60µm and ultraviolet luminosities as
where the luminosities are given by
The values of the UV flux in units of erg s-1cm-2Å-1 and the fluxes at 60µm in Jy are given in Table 1 and the calculated values of the escape fraction 1 - , in Table 2.
The results of including the photon escape are also shown in the UV histograms of Figure 8. As discussed above, under the simple assumption that dust and ionized gas have the same spatial distribution, the values of H and OII do not change with respect to Figure 6.
Figure 8. Histograms of the SFR given by the UV continuum normalized to the SFR(FIR). In the left panel we show the original data with the fixed correction given by Steidel et al. (1999). The central and right panels show the UV corrected for dust extinction computed including the underlying Balmer absorption (see Section 5.1) and photon escape. The median and standard deviation are given for each case. The total number of objects is 25.
In the left panel, the UV continuum was corrected using Steidel et al. (1999) simple approach of multiplying the observed UV flux by a fixed amount ( × 5) which corresponds to the average correction found in a sample of local starburst galaxies [Calzetti et al. 1994] . The fact that the average correction for Calzetti's sample and the average correction for our sample are almost identical suggests that both samples have been drawn from the same family of objects, i.e. they are similar samples.
The central panel shows the result of using the MW extinction law and applying to the observed values the expression A5 where the visual extinction is Av* . The panel at the right shows the result of using the extinction law given by Calzetti et al. (1994). Also in this case the visual extinction is Av*.
Inspection of the UV distribution in figures 2, 6 and 8 shows that at least for the galaxies in the "reference" sample, the escape of photons is a minor effect compared to the underlying Balmer absorption.
Comparing the MW and Calzetti's extinction corrections we can conclude that the corrections using the Calzetti's extinction curve give a SFR(UV) identical inside the errors to the SFR(FIR) plus a substantial reduction in the dispersion of the SFR estimates. These two points justify the use of Calzetti's extinction corrections in samples of starburst galaxies similar to the ones used in this work. It is a remarkable result that the application of our method succeeds in cutting down the scatter present in the original data to less than a half. These two results, i.e. the very close agreement between the SFR(UV) and SFR(FIR) plus the rather small scatter in the ratio of these two estimators gives confidence in the use of our approach to estimate dust extinction and star formation rates in starburst galaxies.