We obtain new age and mass estimates for the star clusters in M82 ’ s fossil starburst region B , based on improved fitting methods .
Our new age estimates confirm the peak in the age histogram attributed to the last tidal encounter with M81 ; we find a peak formation epoch at slightly older ages than previously published , \log ( t _ { peak } / { yr } ) = 9.04 , with a Gaussian \sigma of \Delta \log ( t _ { width } ) = 0.273 .
The actual duration of the burst of cluster formation may have been shorter because uncertainties in the age determinations may have broadened the peak .
Our improved mass estimates confirm that the ( initial ) masses of the M82 B clusters with V \leq 22.5 mag are mostly in the range 10 ^ { 4 } -10 ^ { 6 } M _ { \odot } , with a median mass of M _ { cl } = 1.08 \times 10 ^ { 5 } M _ { \odot } .
The formation history of the observed clusters shows a steady decrease towards older ages .
This indicates that cluster disruption has removed a large fraction of the older clusters . Adopting the expression for the cluster disruption time-scale of \mbox { $t _ { dis } $ } ( M ) = \mbox { $t _ { 4 } ^ { dis } $ } ( M / 10 ^ { 4 } \mbox { $M _ { \odot } $ } ) ^ { \gamma } with \gamma \simeq 0.62 ( Paper I ) , we find that the ratios between the real cluster formation rates in the pre-burst phase ( \log ( t / { yr } ) \geq 9.4 ) , the burst-phase ( 8.4 < \log ( t / { yr } ) < 9.4 ) and the post-burst phase ( \log ( t / { yr } ) \leq 8.4 ) are about 1 : 2 : { 1 \over 40 } .
The formation rate during the burst may have been higher if the actual duration of the burst was shorter than adopted . The mass distribution of the clusters formed during the burst shows a turnover at \log ( M _ { cl } / M _ { \odot } ) \simeq 5.3 which is not caused by selection effects .
This distribution can be explained by cluster formation with an initial power-law mass function of slope \alpha = 2 up to a maximum cluster mass of M _ { max } = 3 \times 10 ^ { 6 } \mbox { $M _ { \odot } $ } , and cluster disruption with a normalisation time-scale \mbox { $t _ { 4 } ^ { dis } $ } / \mbox { $t _ { burst } $ } = ( 3.0 \pm 0.3 ) \times 10 ^ { -2 } .
For a burst age of 1 \times 10 ^ { 9 } yr , we find that the disruption time-scale of a cluster of 10 ^ { 4 } M _ { \odot } is \mbox { $t _ { 4 } ^ { dis } $ } \sim 3 \times 10 ^ { 7 } years , with an uncertainty of approximately a factor of two .
This is the shortest disruption time-scale known in any galaxy .