We analyzed radio and X-ray observations of GRSÂ 1915 + 105 , between May 1995 and June 2006 , focusing on the times characterized by radio flares and cycles of hard dips – soft spikes in the X-ray lightcurve .
Assuming these flares to be discrete ejections , we applied a plasmon model to the radio data , with good agreement with the lightcurves .
We fitted a total of 687 radio flares with a standard model of a plasmon .
We found that the distribution of width is t _ { 0 } = 1160 s with an rms deviation of 360 s , while that of the amplitude is S _ { max } = 59 mJy with an rms deviation of 28 mJy .
The distribution of width is thus rather peaked , while that of the amplitude not .

Regarding radio and X-ray links , this study confirms previous observations on smaller datasets , namely that X-ray cycles of hard dips – soft spikes are always followed by radio flares .
A strong correlation is found between the length of X-ray “ dips ” in the X-ray lightcurves , and the amplitude and fluence of the subsequent radio oscillations .
A model of an exponential rise of the form L _ { \mathrm { 15 } Ê \textrm { } \mathrm { GHz } } ( \Delta t ) = L _ { max } ( 1 - \exp ( - ( \Delta t - \Delta t _ { min } ) / \tau ) is in good agreement with the observations , with the maximum fluence L _ { max } of the order 70 Jy.s , and the characteristic time \tau of the order 200-500 s. We discuss possible physical interpretations of this correlation , regarding the nature of the ejected material , and the physical process responsible for the ejection .