Ultra high energy cosmic ray (UHECR) chemical composition is
important to resolving questions about the locations of UHECR
sources and propagation models. Because composition can only be
deduced by a process of statistical inference via the observation of
air shower maxima ($X_{\mathrm{max}}$), UHECR observatories with large data
collection rates must be employed to reduce statistical
fluctuations. Telescope Array (TA), the largest cosmic ray
observatory in the Northern Hemisphere, is designed to answer the
question of UHECR composition, as well as other important features
of cosmic ray flux, by combining a large array of over 500
scintillation surface detectors spread over 700~km$^{2}$, and three
fluorescence detector stations overlooking the array. With eight
years of data recorded, results of the measurements of UHECR
composition will be presented. UHECR composition is traditionally
measured by comparing the first and second moments of the
distributions of shower maxima, which evolves with energy, between
data and simulations. Reducing statistical fluctuations in the data
helps to distinguish between different primary elements in the
flux. In the current generation of cosmic ray observatories, UHECR
data sets are large enough, and statistical uncertainties are now
small enough, that we can safely distinguish between very light
primary source flux (i.e., protons) and heavy flux (i.e.,
iron). Reducing systematic uncertainties is also important though,
since large systematic shifts in air shower maxima will influence
the interpretation of the data when compared to models. TA therefore
employs different methods of measuring \xm, including stereo air
fluorescence, air fluorescence-surface counter hybrid, and a new
technique using only surface counters. Updated results of TA hybrid
composition among the different methods are presented using up
to eight years of data. Agreement among all TA hybrid composition
results are shown as well as detailed systematic errors which
can be further explored by comparing composition results of the
different measurement methods. Comparison of TA $X_{\mathrm{max}}$ data are
compared to different composition models as well.