diff --git a/analysis.md b/analysis.md
index c749007..1e73b3e 100644
--- a/analysis.md
+++ b/analysis.md
@@ -325,11 +325,9 @@ I plotted curves for both the Weber-Fechner model (log) and the Stevens model
 (pow) with different parameters.
 
 -   The log curve with a flare of 0.05 (WCAG 2) is closer to the pow curve with
-    an exponent of 1/3
+    an exponent of 1/3 and a flare of 0.0025
 -   The log curve with a flare of 0.4 is closer to the pow curves with
     exponents 0.56 and 0.68 (similar to APCA)
--   The pow curve with an exponent of 1/3 **and** a flare of 0.025 is somewhere
-    in the middle.
 
 This shows that a big part of the different results between WCAG 2.x and APCA
 are caused by a different choice in parameters. If we were to change the flare
diff --git a/plots/lightness_comparison.png b/plots/lightness_comparison.png
index 7dda874..fe7afc7 100644
Binary files a/plots/lightness_comparison.png and b/plots/lightness_comparison.png differ
diff --git a/plots/lightness_comparison.py b/plots/lightness_comparison.py
index 3fd297b..13e1cc6 100644
--- a/plots/lightness_comparison.py
+++ b/plots/lightness_comparison.py
@@ -23,8 +23,7 @@ def stevens(flare, alpha):
 
 	weber(0.05)
 	weber(0.4)
-	stevens(0, 0.333)
-	stevens(0.025, 0.333)
+	stevens(0.0025, 0.333)
 	stevens(0, 0.56)
 	stevens(0, 0.68)