This variety is listed in the SG Specialised catalogue as occurring on Pl.36b R20/1 (N14n). Personally I have never considered this warranted its listing; in my opinion it was caused purely by over-inking as it can be found in the same position on a range of controls over a considerable period.
A recent enquiry from a collector of GB overprinted for use in Ireland caused me to look more closely at this flaw. This collector has examples on control T22 which appear to be constant - unfortunately there are no marking on a control block of six but by process of elimination he thinks it is probably from plate 70. However the flaw has been seen on a non-overprinted T22 block with the same control characteristics showing a previously un-recorded plate marking 'dot (top) under E of Penny' on R20/5 but not the recorded Plate 70 markings. Looking at my own control items I see that although I do not have the flaw on any of my T22, I do have examples on controls U22, U23 and V23, none of which have any markings to identify the plate, nor anything to indicate whether or not any came from the same plate. I also have a J17 example from plate 34b, but I have another example of this plate with the same control showing no solid shading although still over-inked. It was on the basis of these two items that I came to be conclusion that this was an inking flaw and only likely to be semi-constant over a short period of the print run. I also have a number of examples from other controls/plates showing considerable over-inking on this stamp although parts of the lines are still visible under magnification.
As the 1/2d design is shared with the 1-1/2d, I decided to look at controls from this value also, and found two examples of solid shading on T 22 plate 26, plus a number of additional controls/plates showing obvious over-inking in this position.