diff --git a/anal.tex b/anal.tex
index 01a7d8a..31c4b10 100644
--- a/anal.tex
+++ b/anal.tex
@@ -153,7 +153,9 @@ \subsection{SiPM noise}
 and correlated noise probabilities in FBK devices similar to the LFoundry
 device we characterize here. The dark count rate, converted from rate per area
 to rate per PDM (area \SI{25}{cm^2}), is \SI{25}{cps}. The correlated noise
-probabilities are: DiCT \SI{17}\%, AP \SI{17}\%, DeCT \SI{1.5}\%.
+probabilities are: DiCT \SI{17}\%, AP \SI{17}\%, DeCT \SI{1.5}\%. These are the
+probabilities for a \SI1{PE} pulse to generate at least one pulse of the
+specified category.
 
 Among correlated noise mechanisms, in the following we focus on DiCT and AP.
 For the DeCT we will just give a very rough estimate, since we are not able to
diff --git a/darkmat.tex b/darkmat.tex
index d549604..d52eacd 100644
--- a/darkmat.tex
+++ b/darkmat.tex
@@ -279,10 +279,10 @@ \section{Direct detection}
 probed by these searches.
 
 The lower limit comes from the mass of the target particle. The scattering
-would be detected by the recoil of the nucleus, which produces ionization and
-scintillation light. If the recoil energy is too small it can not be detected,
-with different thresholds depending on the detection technique, and of course
-the recoil momentum goes to zero as the incoming particle mass decreases.
+would be detected by the recoil of the nucleus, which releases energy in the
+detector. If the recoil energy is too small it can not be detected, with
+different thresholds depending on the detection technique, and of course the
+recoil momentum goes to zero as the incoming particle mass decreases.
 Considering electrons instead of nuclei, lower masses can be probed, but the
 measurements are less sensitive because the radiation background of neutrons,
 impacting nuclei, is kept better under control than gamma and beta rays which
@@ -310,11 +310,11 @@ \section{Direct detection}
 The two most popular choices for the target material are liquid argon and
 xenon. Being fluid, they can be purified from radioactivity to an high degree,
 more than a solid scintillator. Being noble elements, ionized electrons can
-drift without producing a shower or recombining, allowing to use a time
-projection chamber (TPC) to reconstruct the position, which is needed to
-exclude signals near the surface, where there is additional background from the
-surrounding components of the detector. Of the stable noble elements, they have
-the best scintillation yield, about 40 photons per~\si{keV}.
+drift without recombining, allowing to use a time projection chamber (TPC) to
+reconstruct the position, which is needed to exclude signals near the surface,
+where there is additional background from the surrounding components of the
+detector. Of the stable noble elements, they have the best scintillation yield,
+about 40 photons per~\si{keV}.
 
 There are other dark matter models and other kinds of experimental techniques.
 At the large hadron collider (LHC), dark matter is searched through missing
diff --git a/snr.tex b/snr.tex
index 2e3fed1..192c495 100644
--- a/snr.tex
+++ b/snr.tex
@@ -246,7 +246,7 @@ \section{The fingerplot}
 we should place a threshold to discriminate signals, however it is sufficient
 to use any reasonable definition for the purpose of comparing different
 filters. Occasionally we computed the SNR without strictly adhering to the
-definition above, and this will be signaled in the text were appropriate.
+definition above, and this will be signaled in the text where appropriate.
 
 For each event we compute the filter output at a fixed temporal delay from the
 leading edge of the laser trigger pulse, see \autoref{fig:filtersample}.
diff --git a/thesis.pdf b/thesis.pdf
index 0c0d739..0efb326 100644
Binary files a/thesis.pdf and b/thesis.pdf differ