deletions | additions       diff --git a/figures/long_timeseries1/caption.tex b/figures/long_timeseries1/caption.tex  index 85d7ef6..c79b315 100644  --- a/figures/long_timeseries1/caption.tex  +++ b/figures/long_timeseries1/caption.tex 
...
\textbf{Pumping detection at low magnification (4x)}. All panels correspond to $20$ min recordings of wild-type animals in which automated and manual scoring were compared. Unless stated otherwise, the concentration of bacterial food (\textit{E. coli} OP50) was $OD_{600}=3$. $OD_{600}=2.5$.  In panels A-C, automated and manual scoring are plotted as red and blue lines, respectively. (A) An example of the cumulative number of pumps as a function of time ($R^2 = 0.98$). (B) The data from panel A binned in 30 sec bins ($R^2 = 0.90$). (C) Distribution An example of a distribution  of instantaneous pumping rates (intervals between consecutive pumps). Inset: correlation between the automatically detected and manually detected pump events. The dashed line denotes unity. Intensity corresponds to the density of pumping events. (D) Mean pumping rates from $5$ animals (one per data point). Here, the instantaneous pumping rates were smoothed with a median filter of width $15$. The dashed line denotes unity, the major and minor axes of the ellipsoids represent the standard deviations of the two rates. To obtain a range of mean rates, we assayed $5$ animals at food concentrations between $OD_{600}=3$ and $OD_{600}=5$ and scored $500$ pumping events per animal. \label{fig:longtimeseries}