It is well known that ultrasonic processing is a crucial step for preparing suspensions of individualized single-walled carbon nanotubes (SWCNTs). Sonication disaggregates nanotube bundles but can also cut long nanotubes into shorter segments and introduce defect sites that degrade optical properties. To study these processes in detail, we have applied the newly developed method of variance spectroscopy.1 In this technique, we analyze fluorescence spectra from many different small regions of a dilute SWCNT sample to determine the variations in emission brought about by statistical changes in composition. By analyzing the first and second moments of the intensity distributions at each wavelength, we construct the mean and variance spectra. Simple analysis of these spectra then reveals the number of particles per unit volume of different (n,m) species present within the mixed sample and the average emission per particle for various (n,m) species. Deeper data analysis gives covariance plots, showing aggregation between different species, and third moment (skew) spectra, showing aggregation of individual species. Starting with a sample of relatively long SWCNTs that had been dispersed in surfactant solution by high shear mixing, we measured variance spectra as a function of sonication time and power. We found that the number density values for every (n,m) species increased with sonication time, reflecting initial further disaggregation and a slower cutting process. These rates of number density increase showed no apparent dependence on nanotube diameter or chiral angle. To confirm the changes in SWCNT length caused by cutting, we also measured full length distributions using the diffusion-based LAND method.2 Findings from these length distributions were compared to the change in number density values in order to disentangle the effects of debundling and cutting. We believe that this is the first study to separately observe these effects. Damage to SWCNTs during continued sonication was monitored through decreases in the values of emission per particle, as determined from variance spectra. We also introduce here the use of skew analysis to obtain a quantitative index of self-aggregation. The experimental results will be compared to findings from a numerical model constructed to represent the main processes occurring during sonication. Our study should allow the optimized preparation of SWCNT dispersions. 1. J. K. Streit, S. M. Bachilo, S. R. Sanchez, C.-W. Lin, and R. B. Weisman, J. Phys. Chem. Lett., 3976-3981 (2015). 2. J. K. Streit, S. M. Bachilo, A. V. Naumov, C. Khripin, M. Zheng, and R. B. Weisman, ACS Nano, 6 (9), 8424-8431 (2012).