diff --git a/README.md b/README.md
index a19b6a701f3359a99c7de280950e1d527e408890..0429aa29ff28f67a862d85f842a5cb3c5e5dd8fe 100644
--- a/README.md
+++ b/README.md
@@ -48,7 +48,7 @@ In the case of a coil, center axis of the coil is at a fixed distance from the c
 
 ## Differences between this and a commercial FEA package
 
-* Erik pointed out that rather than go stright to B fields, most commercial solvers will likely calculate H and then...
+* Erik pointed out that rather than go stright to B fields, most commercial solvers will likely calculate H and then... This explains why commercial solvers are so concerned with enforcing an infinite air boundary around their models...
 * Real FEA packages do a lot of front-end work to mesh their models (triangular meshes rather than just enforcing a uniform grid). This allows for users to get higher resolution results in regions of interest, and avoid enforcing an element size dictated by the simulations smallest regions of interest. With that said, the meshing problem can be looked at as completely seperate from the physics, and I'd imagine at Comsol there is a meshing department and then various physics departments that don't interact too much with one another...
 * Run on the GPU