There exists another XMM-Newton observation (obtained on 12 April 2003), which however was strongly affected by a high, flaring radiation background and was therefore not used in our analysis; inspection of these data showed that none of the class I protostars discussed in this paper was detected in that observation. of the Serpens cloud that were obtained between April and September 2003. These three data sets were not affected by strong background variability, and their full exposure times could be use for scientific analysis.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

It is interesting to note that the detection fractions among the more evolved YSOs in Serpens is even lower: just two (i.e. 15%) of the 13 flat spectrum objects from Kaas et al. (2004) and only four (i.e. 9%) of the 43 class II T Tauri stars listed by Kaas et al. (2004) are detected in X-rays. One reason for this relatively low X-ray detection fractions is that the sample of cluster members was identified in very deep infrared observations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... $\sim$ 13 h

We assume two flares for EC 89 and two flares for PS 1, i.e. four flare-like events seen on four objects observed for 12.8 h. Note that all our X-ray detected class I protostars would also have been detected in the merged data set without their flares; this implies that our flare frequency is not overestimated by inclusion of objects that were only detected due to a flare.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.