Log-Normal Distributions in Gamma-Ray Burst Time Histories

We propose a new, simple but powerful algorithm to analyze the gamma-ray burst temporal structures based on identifying non-statistical variations (``peaks'') in the time histories. Detailed analyses of the bursts from the third BATSE catalog show that $\sim 30$ bursts have more than 20 peaks individually. Upon identifying most of the peaks in those bursts, we show that the peak fluence $S_i$ and peak interval $\delta_i$ distributions within each burst are consistent with log-normal distributions. Furthermore, we show that Gaussian (in linear space) and power-law distributions for peak fluences are ruled out, as is the Poisson distribution for peak intervals.