FIGURE 2. Evolution of the IR flash associated with GRB 041219a.

We argue that the initial pulse is not traditional afterglow from an external shock but instead related to the internal shocks of the central engine. Shown (top three panels) are measurements derived from individual 7.8-s exposures as small, light-coloured points, and detections from stacks of images as heavy, dark-coloured points; error bars are 1, estimated from photon noise and the distribution of randomly placed apertures on the images. Also plotted (bottom panel) is the light curve from the 15−350-keV channels of the BAT instrument aboard Swift. Data reduction was as follows: the response to variable sky and bias in the detectors was estimated for each exposure in each band by median-combining all the exposures taken within a two to four minute window. A flat-field correction for fixed pixel-to-pixel variations in detector gain was made using images of the dawn sky. Subtracting the dark + sky response and normalizing by the flat-field produced a reduced image. Reduced images were measured either individually or as stacks of co-added images. As individual images undersample the seeing owing to atmospheric blurring, photometry was performed differentially in an aperture of fixed size. All of the images were aligned to a common reference image to an accuracy of approximately 0.1 pixel. Images were measured individually or in stacks created by summing individual images with weights determined by the signal-to-noise of each image. The measured flux at the position of the GRB was compared to the flux measured for a set of nearby comparison stars. Magnitudes of the comparison stars are known to a high accuracy (2%) from the 2MASS catalogue. PAIRITEL uses the same detectors and filters as 2MASS, so our differential photometry is expected to be free from systematic offsets.