rolling.html: Note that the results are unreliable under some circumstances.

diff --git a/rolling.html b/rolling.html
index 14e337f..1feb248 100644 (file)
--- a/rolling.html
+++ b/rolling.html
@@ -173,6 +173,11 @@ of it you need to create the required starting size that you can then
 roll down to the required thickness of strip.
 
 <p>For best results, roll the strip in as few passes as you can handle.
+The results aren&rsquo;t accurate for broad, thin strips.
+Width-to-thickness ratios of up to 6:1 are known to be good; ratios of
+12:1 or more don't work: strips don&rsquo;t get as wide as predicted by
+our model.  We need to collect more data and construct a better model to
+handle this case properly.
 
 <p>You can see the <a href="rolling-eqn.html">detailed equations</a>
 used for this calculation if you&rsquo;re interested.