deletions | additions       diff --git a/untitled.tex b/untitled.tex  index d4408a7..4e3e3b6 100644  --- a/untitled.tex  +++ b/untitled.tex 
...
We also apply a correction for mass loss to both the local group and \cite{Whitaker_2014} data. As these both determine mass by integrating the star formation rate over time, the data shows the total stellar mass formed by a certain time, rather than the total stellar mass present at that time. Much of this mass loss is caused by the death of high mass, short lifespan ($ < 100Myr$) stars and so can be approximated as instantaneous using a multiplicative factor. The \cite{Tomczak_2014} determines mass from observed luminosity and not star formation rates, and so this is not applied there.  Finally, a morphological environmental  correction is was  applied to the local group data. Approximately 50\% of If we split  the knowngalaxies in the  local groupappear in this data set. However, if we divide these  galaxies into those attached to groups based on their location - satellites of  the milky way, those attached to Milky Way, satellites of  M31 (Andromeda) and those in the field, attached to neither of these two -  we find that we do the \cite{Weisz_2014} data, which contains only a subset of approximately half of all galaxies, does  not sample evenly  from these groups evenly. This correction weights three groups. To correct for this, we weight galaxies to ensure that at  each galaxy mass and redshift the different environments are correctly weighted.  A significant analysis was also performed on the errors on the local group data reported by \cite{Weisz_2014}. These errors were calculated using methods defined in \cite{Dolphin_2012} (systematic) and \cite{Dolphin_2013} (random) but are considered extremely conservative. A method  to account for this sub sampling determine a more reasonable set of uncertainties was not found and so we adopt the convention used in a later paper by Weisz,  The main analysis of the local group data was to do with the error bars reported by \cite{Weisz_2014}. These errors follow the conventions specified in \cite{Dolphin_2012} for systematic uncertainties and \cite{Dolphin_2013} for random. However, these uncertainties are extremely conservative and so instead we use the same convention as in \cite{Weisz_2014} which conservatively a relative uncertainty of 50\%.