**2. GRB Classification and Properties**

To achieve a recognized GRB classification, the strategy is to take into account the most relevant astronomical properties of such objects. Thus, as the most prominent GRB component is represented by the prompt *γ*-ray emission, it is straightforward to use it to define GRB classes based on similarity criteria.

The prompt *γ*-ray emission is characterized by highly-variable and multi-peaked light curves composed of either overlapping or distinct pulses with variable duration. The duration of these pulses spreads within a wide time range. Since the duration is not fixed a priori, it is natural to wonder whether one can arbitrarily define a time in which the above measures can be obtained. Hence, it is a consolidated convention to take the total burst duration in a time interval, dubbed *t*90, evaluated in the observer frame over which the 90%, from 5% to 95%, of the total background-subtracted counts are experimentally detectable.

In view of such a property, one gets a plausible classification, as we report below.

#### *2.1. Classification: Short and Long GRBs*

The light curve analysis of the first BATSE GRB catalogue showed a clear bimodal distribution of the *t*<sup>90</sup> duration, separated at roughly 2 s, and in the hardness ratio (HR), namely the ratio of the total counts of the hard 100–300 keV energy band over the softer 50–100 keV band [3,14,15].

This leads to the widely-adopted classification into


The significance of such a classification scheme has been strengthened with the full 2704 GRBs detected by BATSE and later GRB missions, providing strong evidence for two GRB progenitor channels (see, e.g., Figure 1).

**Figure 1.** GRB distribution provided by the first BATSE catalog, lying on the *t*90–HR plane. The solid HR histogram shows LGRBs, whereas the dotted one is for SGRBs. The dashed horizontal lines mark mean HRs for both classes. The solid *t*<sup>90</sup> histogram represents the raw data whilst the dotted one shows the error-convolved data, credit from Ref. [3].

However, a significant overlap in the distributions of SGRBs and LGRBs suggests that *a more robust classification scheme based on physical properties is still missing.*
