diff --git a/docs/rendering/animation-systems.md b/docs/rendering/animation-systems.md
index 6d19f74..284541b 100644
--- a/docs/rendering/animation-systems.md
+++ b/docs/rendering/animation-systems.md
@@ -23,11 +23,13 @@ FrameDisplayTimeInMilliseconds = AnimationClip.FrameTimeInIntervals * 25;
 Before diving into the accuracy of this estimate (_spoiler alert - it's slightly off_), some questions immediately spring to mind:
 
 - Why is the frame time given in "intervals" (as ACT editor calls them) in the first place?
-- Where does the 25 ms time value that has so far been used by the community actually come from?
+- Where does the 25 ms time value that has so far been used by the community actually come from? [^1]
 - Is it at all correct, and is there a way to precisely measure sprite animation times to verify/disprove this?
 
 The following sections attempt to answer these questions, as far as is realistically possible.
 
+[^1]: The earliest known source appears to be [an earlier specification of the ACT format](https://web.archive.org/web/20200220130616/http://mist.in/gratia/ro/spr/ActFileFormatFix.html), which indicates a tool called actOR
+
 ### Animation State Updates
 
 In order to answer the first question, a bit of guesswork is needed. There's of course no way of telling for sure why this peculiar unit of measurement was chosen by the game's developers. But knowing a bit about the different animation systems in the client, a likely explanation is that sprite animations use a different mechanism altogether (e.g., clock-based state machines).