Assuming you did walk 400m in a straight line, and you ended up 85m to one side* (perpendicular to your intended track), that gives a bearing error of 12 degrees. You should certainly be aiming for better than that, and, ideally, close to the bearing error of your compass. Otherwise, use smaller legs, or use 'aiming off' technique to help you get to your leg target.
If your pace count is 44 per 100m, that means each leg stride is 1.14m (and each pace 2.28m). That's quite a stride; do you have very long legs? Note Lyle's comment in the book ('PC typically varies from 55, for very tall people, to 75 for people with short legs').
How did you determine the desired bearing? What technique did you use to maintain the bearing? You say you took a bearing to a track; could your aiming point be uniquely identified (linear features such as tracks can be hit at any point along their length, so usually require a secondary reference to provide a resection to give a unique position along the track)? Could it be that your bearing was taken to a point on the track that was different to the grid ref you thought it was?
Are you sure your GPS readings were sufficiently accurate? Were there any hills or buildings that may have perturbed the received signals? Was there any tree cover? Had you let the receiver acquire a good constellation; you say you turned your GPS on, and the position can take a while to settle as additional satellites are acquired and added to the position solution.
* or did you end up 85m short of your target? In which case, this would give a PC of 100*176/(400-85) = 56, which might be a bit more reasonable.
As you can see, there are many ways in which measuring and following a bearing can go wrong! Good for you for getting out and practising.