MDM,
the most likely cause is scope adjuster error. Very few scopes, even top end ones are within 5% accuracy on a 'round the box' test at 100yd in settled conditions off the bench with a rifle and load capable of 0.25-MOA or smaller groups. That is, set the scope to shoot at the bottom corner of a large sheet of paper, shoot one, add 20 or 30-MOA and shoot another one; go right the same amount in MOA etc, then down etc, then left etc.
Ideally when finished, shot 5 touches shot 1 or is very close, the four corner holes produce a true square with 90-degree angles and equal length sides and the length of the sides is the scope adjustment values X 1.047". In real life the error is usually on the plus side giving sides to the square that are greater than the recorded MOA value in inches.
A friend has run this test many times using his Stolle 6PPC LV gun in scope reviews for publication with Leupolds, S&Bs, Nightforces, Sightrons, and believe me a mere 5% error in this test is small, many expensive target scopes having MOA values nearer to being 10% out from the recorded values. On the other hand, consistency of values has improved greatly in recent years so he does tend to get a true square with recent production top end models.
So, take a 5-10% scope adjuster error, a small MV chronograph error, different atmospheric conditions from those run through the ballistic program - and as Bryan Litz is the first to point out - small errors in the experimentally derived G7 BC values, and a 2-MOA 'error' between estimated and required come-up at 900 or 1,000yd from a 100yd zero is very easily obtained unless the 'errors' luckily cancel each other out. The objective must always be to get the bullet somewhere onto the target frame in such conditions and it's very optimistic to believe that a precise result can be predicted in advance of the shot no matter how good the ballistics program, atmospheric metering gear used on the range or in the field etc for such a large range change between two shots.
The good news from my friend's tests is that the adjuster error issue is being addressed by the scope manufacturers as a 'mere' 5% error would have been regarded as fantastic when he started this 10 or 12 years ago while the best of today's models like the S&B 12-50X56 PMII and Sightron Series III L-R scopes are now achieving it.
Laurie,
York, England
