diff --git a/README.md b/README.md
index 23d3923b994bcbf5e3e6cb15e74dc2cbee4ffdc2..7f94108e463ffd135eee53a27e4b296cb1d98a85 100644
--- a/README.md
+++ b/README.md
@@ -597,6 +597,12 @@ print("  p-value = %g" % p)
 
 ```
 
+The additional `-1` stems from the fact that we have a Poisson distribution
+for which the uncertainties are derived from the number of events. The
+test is not sensitive to the normalization, and therefore, we loose one degree
+of freedom, cf. [Data Analysis in High Energy
+Physics](https://www.wiley.com/en-us/Data+Analysis+in+High+Energy+Physics%3A+A+Practical+Guide+to+Statistical+Methods-p-9783527410583) by Behnke et at.
+
 If you run the `decay.py` you should see the following fit results.
 <!-- console_output -->
 ```bash