The 2d Blue Problem

A Possible Solution To The Trapezoidal Appearance

Much has been made recently of the appearance of the 2d Blue stamps from Plate 3 being 'trapezoidal' in nature. Measuring the stamps agrees with this, being that the majority of the impressions are about seven hundredths of a mm larger in height on the left side.

In this block from Plate 3, measurements taken close to the left and right frame lines show that the left side is consistently larger than the right. While the overall size of the transfer may vary up to 0.13 mm the difference of left to right remains about 0.07 mm. Variance from this are probably due to the inking and measuring error.

Measuring was done using a 2400 PPI scan of the block on an Epson 3200 Photo scanner. Ideally it should be done at 2540 PPI, being 1/100th of a mm to a pixel. Using the scanner's Optical resolution and then calculating the adjustment based on a 1.058333 relationship between optical and desired allows measuring to the hundredth mm.

This keeps the measuring based on actual scanned and viewed pixels, not created and averaged pixels as would be obtained by using a sampling algorithm within the scanner's software.

In addition, again to avoid pixel manipulation, the scans were repeatedly taken until the image obtained was properly 'level' for measuring. Photoshop allows holding the SHIFT key to guarantee a vertical measure of the Ruler Tool.

Having established that physically the 2d stamps of Plate 3 are indeed longer on the left side than on the right, what is the cause for this and are other 2d stamps so affected? Is this a problem from the laying down of the plate, from taking up the roller impression or where may the fault be laid? Are any of the 1d Line-Engraved stamps also affected?

Beginning with full margin multiples from the Penny Reds, Plates 10, 13, 18 … and on through the remainder of the Imperforate period, there is no consistent difference between the sides on the impressions. Such measured differences as found come with a retouched, weakened or scraped corner frame.

At this point it would seem that we may eliminate the master ONE PENNY die and the laying down of the plates, as certainly either should have had a similar affect upon at least some of the Penny Red stamps.

Perhaps something in taking up an impression on the 2d transfer roller could be to blame? If so, then this situation should be limited to the plate or plates made from this single roller impression.

What is the situation for Plate 4 … for Plate 5 … and for the following plates to Plate 6?

In this block from Plate 4 we see the same situation, the left side is consistently larger than the right. Additional blocks and strips of Plate 4 from the author's collection confirm it happening throughout the sheet.

Plate 5 was made from the same roller impression used for Plate 4, thus this pair shows the same relationship between the left and right side sizes.

Single impressions from Plate 6 continue this arrangement, but here there is no sign of the characteristic weak “8 o'clock Ray Flaw” from Plates 4 and 5. Other studies concluded that a fresh roller impression was used for Plate 6, yet the relationship between the left and right sides persists.

We know the transfer roller impressions were modified to produce the “White Lines” and that at least three such roller impressions were used. Plate 3 seems to stand alone, Plates 4 and 5 have a common flaw and Plate 6 is the last of the Star plates.

So far we've worked forward in time from the original finding of 'trapezoidal' impressions on Plate 3 … but what about the original Plates 1 and 2?

Presented are two examples from Plate 1 of the TWO PENNY plate. Both show the same trapezoidal relationship between the left and right side sizes.

At this point some mechanical problem with the transfer press can be ruled out as none of the PENNY RED stamps examined have this trapezoidal appearance.

As this affects all of the 2d Star plates, there must be something in the Die or transfer roller used to create these first six plates.

In his “Line-Engraved Postage Stamps” Sir E. D. Bacon writes on page 87;
“The die for the Two Pence stamp was made by taking a roller impression from the die of the One Penny.”

One of the existing impressions not yet used for laying out a plate of the 1d value could have been used rather than taking up an entirely new roller impression. A fresh roller impression would of course be wanted for the best possible transfer.

With the foregoing measuring work at hand and noting that only the 2d values were affected, the roller impressions are probably not the source of the trapezoidal effect. We have now eliminated the laying out of the plates and most probably the transfer roller when making the new Die or on plate layout.

Mr. Bacon then notes;
“Mr. Petch said that the background had been 'Re-Entered' but he evidently did not use this term in its strict sense, as the background could not have been re-entered by a second impression of the roller without touching and possibly injuring the head. What was really done was that the background was carefully gone over by an engraver and the lines deepened.”

A little later in this paragraph, he continues with;
“The deepening of the lines of the crosses in the upper corners and the prolongation of some of the lines is well marked.”

I believe that not only were the background and the crosses deepened, but the corner frame lines were also deepened. In doing so, the engraver did not manage to keep the upper left, upper right and the lower left corner frame lines parallel to the rest of the top and base of the stamp design.

Taking these same stamps and using Photoshop's ability to show degree of the ruler's measuring angle, I find that on actual stamps the upper left corner frame line deviates upwards by 0.30 to 0.70 degrees from the average of the impression above POSTAGE. The lower left corner frame line shows a similar 0.50 to 0.80 degree depression compared to the average of the bottom of the stamp under TWO PENCE. These measurements are very subjective to the inking and wiping of the plate or die in taking an impression. The resultant height also varies when the plate impressions have a weakened NNE line as in Plate 1.

The combination of these deviations agree well with the measured and viewed difference between the left and right sides of the Two Pence stamps. On this Die Proof, [with only a 300 PPI resolution image available to me] the left measures 28.53 mm with the right measuring 28.44.

Previous writings regarding the 'trapezoidal' appearance of Plate 3 of the 2d value now appears to be technically confirmed, when measured along the edge of the impressions. This investigation now expands this 'trapezoidal effect' to all of the 2d Star Plates.

However ...

IF we move the measuring points into the stamp at the juncture of the corner squares with the body of the tablets, then the measurements of the 'sides' becomes much more equal ,within tolerances of scanning and computer measuring.

As an example, the block of six from Plate 3 had the impression “KK” at 22.72 mm vs. 22.68 mm. Moving the measuring points to the inside of the corner squares, we have 22.68 mm vs. 22.67 mm. The 'long' unit “ML” at 22.83 vs. 22.77 becomes 22.74 mm vs. 22.72.

This leads me to a conclusion that the whole of the matter rests with not with any function of the transfer roller or press, but with the actual Die made for the Two Pence stamp 'as retouched' by the engraver. It does not appear to change with uneven shrinkage of the paper as some have suggested.

This conjecture should be fully proven to a higher degree of measurement accuracy if a very high resolution scan of one of the few Two Pence Die 1 Proofs can be obtained. The current degree measurements above were made using a low resolution image of one of the die Proofs. With higher resolution, both the apparent size difference and the verification of the skewed top/bottom frame lines in the corners can be more accurately measured and thus firmly attribute the cause of the 'trapezoidal effect' directly to the original Die made for the Two Pence value.

John A. McCulloch

Created 15th May, 2015
Updated 24th December, 2015