Many thanks for all of the replies.
Is your work piece exactly at the right dimension and did you reference off the same edge of the workpiece?
Yes, I am aware that the workpiece needs to be exactly square to avoid errors in the hole positions. Yes I did reference off the same edge of the workpiece.
did you use the centering mandrel to align the router properly with the mounting plate?
Yes, I did and there was no play at all. Before tightening the screws up hard I tested for possible movement and the only possible movement was rotational which wouldn't affect the accuracy in the hole positions.
Check the space between the stop post and the threaded hole on both stops.
What I actually did check here was (using a digital micrometer) the distance between the end of the stop and the pin (or one of them) that protrudes into the hole in the rail. I got the same measurement with both stops (to within 0.1mm and I doubt if my measurement accuracy is better than this as it is difficult to make this particular measurement). I didn't perceive much play of the stops in the holes of the guide rail but what I do find somewhat unsatisfactory is that whilst you can get the stop to seat well on the rail, when you tighten up the screw, there is a tendency for the stop to cant over because it is being tightened from one side only. This would tend to move the holes nearer to the reference edge of the workpiece. I have taken to tightening up this screw only gently so that I can make sure that when I position the rail the stop is butted up properly and at the same time, lying flat against the rail. I felt this should work as there seems to be no play of the stop against the rail when it is tight up against it and flat.
Let's say you're standing at your work bench with your guide rail parallel to your stomach and the rubber edge facing away from you.
sit your LR32 on the rail in front of you. With the pin in one of the holes on the rail, draw a mental line square off the rail connecting the pin to the large hole in which you're centering mandrel fits
Let's say that the true center of the hole is a quarter of a millimeter to the right of your mental line.
Then you would have 32.25 mm measuring off the left, and 31.75 mm measuring from the right.
Now if your tape measure is reading long by a quarter of a millimeter, you would be reading 32 mm off the left side and 32.5 mm off the right side.
When I check the distance of my holes from the reference edge, I am using the same ruler off the same reference edge to check the two lines of holes so if the ruler had an error, it would have the same effect on both sets of measurements so that any difference in the two sets would reflect a genuine difference in hole positions. Your scenario in which you describe an error in the alignment of the centre of the 'big hole' and where the guide plate perceives the centre of each 'little hole' (I use this description to allow for the possibility that the retractable pin mechanism may not be aligned properly) would introduce errors in hole positions in opposite directions depending on whether I am indexing from the right-hand side of the rail or the left. The error would be the same in each case and the total error in hole positions (one row compared to the other) would be double this error.
If I am not mistaken, your idea of a possible cause involves making the measurement (with a dodgy rule) from opposite edges of the workpiece?
the problem is the result of how the retractable pin fits the holes in the rail.
Thats what I think the problem is.
I will do an experiment to make absolutely sure I can isolate the effect of any slack.
Thanks again for your efforts to help me out. I'll report back over the weekend.