Chains of type I stormbursts

From radio spectra between 160 and 320 MHz of chains of type I bursts it appears that their duration distributions allow an exponential fit, and that those of samples containing long and short chains respectively, taken from the same storm, have virtually the same characteristic time (logarithmic slope). On the average this figure decreases - as a function of the frequency - at about 1 s per 10 MHz. The high frequency cut-off of chain activity (noise storms) is mainly a consequence of the frequency dependence of the probability for the first burst of a chain to appear. Given the density of type I bursts in a chain, it is concluded that the probability of a type I burst to be followed by another one is at least 90% below 250 MHz and 70–80% at 300 MHz, which makes it essential for type I theories to include a mechanism to this effect. The drift rate distribution for chains is symmetrical with a peak at-10 MHz/s. The statistics is indicative of a correlation between drift rate and duration. No evidence has been found for the occurrence of chain pairs or frequency splitting in chains, nor for an association between chains and type III bursts.