In the discussion above and elsewhere in this work we have ignored the effects of trailing loss—when objects move during the exposure time they leave ‘trails’ on the image rather than point sources and the trails have, in the past, been more difficult to identify than point sources. While the PS1 point source detection limit is expected to be V 22.7 the efficiency of our faint trail detection algorithm is not yet known. It is thought that it should be efficient to per-pixel S/N1 implying that the detection efficiency could be quite good even for intrinsically faint objects. Since we have not yet measured the trailing losses we ignore them. This means that the small impactor detection efficiencies reported here and in Fig. 12 are upper limits.