A PRINCIPAL COMPONENT ANALYSIS OF THE 3B GAMMA-RAY BURST DATA

We have carried out a principal component analysis for 625 gamma-ray b ursts in the BATSE 3B catalog for which nonzero values exist for the n ine measured variables. This shows that only two out of the three basi c quantities of duration, peak flux, and fluence are independent, even if this relation is strongly affected by instrumental effects, and th ese two account for 91.6% of the total information content. The next m ost important variable is the fluence in the fourth energy channel (at energies above 320 keV). This has a larger variance and is less corre lated with the fluences in the remaining three channels than the latte r correlate among themselves. Thus a separate consideration of the fou rth channel and an increased attention paid to the related hardness ra tio H43 appear useful for future studies. The analysis gives the weigh ts for the individual measurements needed to define a single duration, peak flux, and fluence. It also shows that, in logarithmic variables, the hardness ratio H32 is significantly correlated with peak flux, wh ile H43 is significantly anticorrelated with peak flux. The principal component analysis provides a potentially useful tool for estimating t he improvement in information content to be achieved by considering al ternative variables or for performing various corrections on available measurements.