Our established understanding of Transfer Roller Lines (TRLs) primarily come from an article by Mr A.G. Brewer in 1972 (Gibbons Stamp Monthly, October 1972), a pdf copy of which can be downloaded here. I have to say that there are some aspects of this article that I disagree with, especially the cause of TRLs. However you should know that two very prominent philatelists have disagreed with me and side with Brewer. Being a bloody minded so and so, that sort of thing has never stopped me from shouting out any way. You should proceed with caution. The following quotes are extracts from Mr Brewer's article.
“...a clearly defined linear mark as on this plate, is in fact caused by the edge of the Transfer Roller, which, when applied to a non-softened plate on the small Transfer Press, rocked unevenly or was not held at a ninety degree angle, and so impressed a chamfered edge into the plate, clear of the stamp unit (Fig. 1).”
Brewer's original drawing.
I would disagree with this theory on two main points.
Firstly, I would suggest that the axle for the transfer roller would have been absolutely level. There would have been no reason for it to be otherwise. Am I alone in thinking that the phrases “thousands of pounds of pressure” and “Careful, it wobbles a bit” would probably not be uttered together in the same breath?
In addittion, we have examples of TRLs appearing on both sides of a repaired impression. Now it is conceivable for a roller to be out of square one way or another, but not for it to wobble from side to side. Consider this rather nice strip from state 2 of plate 17 [courtesy Kevin Maunder].
Starting from the left, we can see a feint TRL to the left of CC. The margin between CC and CD shows two TRLs. There appears to be but one TRL between CD and CE but I would argue that it is possible for the two TRLs to be coincident as this margin is narrower than it's neighbors. The margin between CE and CF again shows two TRLs, and I can just make out a feint TRL to the right of CF. This is inconsistent with a roller that is positioned at an angle to the surface of a printing plate. Other examples can be found of later states with TRLs both side of the stamp (e.g. 54 SH; 2d blue plate 3 QC).
Looking at the imp sheet for plate 17, it would appear that the impressions are fairly well balanced, with the left side of the impressions being only slightly stronger than the right. On most repaired plates, the weaker side tends to get much firmer, and that is the side that the TRL is likely to appear. The obvious conclusion to this is that there is more metal being displaced on that side of the stamp, which leads to more trauma to the plate. However on plate 17, the metal has flowed both left and right more or less equally.
Secondly, going below the plate surface level is anathema to an engraver. The aim is to impart the design on the roller into the plate - and nothing more. Anything else would have been visible after printing, indeed, taking the design below the mean level of the plate would have resulted in an inky puddle that would resist attempts at wiping the excess ink. In short, the plate would have been ruined. Moreover, setting the limit of vertical travel of the roller would have been relatively easy for the engraver to achieve, by the simple expedient of resting a blank portion of the roller on the plate surface and tightening the respective locking nuts. Care! I do not know this as fact and have not been able to identify any kind of mechanical stop from illustrations of the presses used. However an engraver works in three dimensions, not two. I would argue that this would have been one of the fundamental setting up steps taken with any engraving job. A locking nut is a very simple mechanism to incorporate.
“It must be noted that the side line of the printed stamp is invariably stronger on such impressions, confirming greater pressure to one side during the transfer of the to-be re-entered unit."
It is true that when the side of an impression has been strengthened, the TRL tends to be evident on the same side, but that does not necessarily mean that the roller was at an angle to the plate surface. Only that that is where the most trauma is occurring and a greater amount of metal is being displaced. It is not uncommon for one side of an un-repaired stamp to be stronger than the other, the difference probably being due to a slight distortion of the roller during the hardening process. On new plates that show this feature, it is clear that the roller edge was not allowed to dig into the plate surface to try and compensate, since this would have been more troublesome than having one side of the impression weaker than the other. In these later cases, the side network was normally strengthened by re-cutting the side by hand, or just left as they were.
“Note: In the case of impressions which have to be repaired by re-entry (or fresh entry)... the repaired impression will almost certainly be fractionally lower than its unrepaired adjacent impression, and in the case of second or third repairs by re-entry on the same unit the difference in levels may be even more pronounced. As the Transfer Roller is slightly wider than the actual stamp design the tendency for the 'dressed' edges to cause an incursion into the metal of the marginal gutter increases with increase in the relative depth of penetration. (Re-entered unit to actual plate unit levels.) Perhaps in some cases of a second re-entry the remaining burr (which becomes increasingly difficult to remove without affecting the stamp design) tends to make this incursion even more marked.”
The idea that in order for a re-entry to be successful the roller has to go below the level of the plate is incorrect. The aim of re-entry is to deepen the engraved lines relative to the mean plate level. That would mean that the engraver would stop rolling in the impression as soon as the roller came level with the plate surface, the work of deepening the impression being done by the raised relief on the roller surface. He certainly would not want to have the impression at the bottom of a furrow on the plate. It is worth trying to imagine the difficulty of wiping excess ink from a plate that has such a depression, which would make the printing operation difficult to impossible.
164 TD state 2. Courtesy Scott Treacey
Just about as deep as a re-entry gets!
"That the 'kink' is in fact an irregularity on the edge of the Transfer Roller, possibly caused when the burrs, made at the time of taking up the die, were cleaned off...”
The kink that Mr Brewer describes seems to correspond with the blank upper corners of the design. That would be consistent with metal displacement as the largely flat corners would mean that less metal needed to be displaced when compared with the body of the stamp. Suggesting that it corresponds with an irregularity on the edge of the roller, is no more than guesswork. Any shaving down of the sharp edge of the roller would have been completed on a lathe and the chamfer would have been completely uniform. Kevin's strip from plate 17 also shows a very slight kink at the base of the TRLs between CC-CD and CE-CF. And this is clearly from a different transfer roller.
So what is actually happening during re-entry?
In a new entry on a soft, blank plate, the hardened roller is forced into the metal until the surface of the roller is level with the surface of the plate. The metal that is displaced bulges out the sides of the roller as a burr. A smooth bulge would be normal as the metal (ferrite) is at this stage comparatively soft and homogeneous. This metal bulge is removed with a burin so that the area is made level with the plate surface.
Once all of the cleaning up is completed, the plate is normally put through a hardening process, which sees the plate placed in an iron box closely packed with carbon rich material and heated in a furnace. After quenching, the metal forming the surface of the plate completes a structural transformation. The steel changes from reasonably homogeneous ferrite into a ferrite/cementite/ bainite/pearlite mix. In the mid 1800s, this process was more of an art than a science, so the exact composition and depth of the hardened layer was very variable. But the end result was that the plate had a hard skin. An analogy could be the skin on a rice pudding that your granny made (not some bloody ghastly stuff that comes in a can. If that forms a skin, it's sentient).
Brewer's drawing reworked.
Here is my badly drawn attempt to show what I think happens before and during re-entry, with all effects being grossly exaggerated.
Figure A shows the process of a roller entering an impression into a soft ferrite plate. The bulge (burr is really the wrong word to use) to the left of the roller will be removed during the cleaning up process, so that the mean plate level is flat.
Figure B shows the condition of the plate after it had gone through the case-hardening process. The shaded area of the drawing represents the areas of the plate that have made a structural change. The metal below this level will be soft ferrite. One problem with the rice pudding skin analogy is that it infers a strong hard - soft demarcation line. On a printing plate, the transition would be gradual, with less bainite etc found, the further into the plate you go.
Figure C shows the condition of the plate after it had been at press for some time and has suffered a degree of wear. The dotted line represents the metal that has been worn away. To the right the transfer roller has been positioned in the remaining impression left on the plate, but the rolling in process has yet to start. It is worth making a couple of points at this juncture. It will be seen that at the bottom of the impression the layer of case hardening is still at its original depth, since mechanical plate wear has occurred at the plate surface.
Figure D. It will be remembered that case-hardened metal has a physical structure which would have very different properties to soft steel and which when displaced would react differently. Areas beneath the mean level of the plate will still have a thick layer of case hardening in place, and as the roller will start to act directly here before any other area, it is likely that the hardened metal will take the line of least resistance. That is, down into the softer ferrite beneath the harden layer. This displaced metal has to go somewhere; upwards in central areas of the stamp, damaging the hard metal structure in the process, and; sideways flowing away from the stamp into the margin. As the process is continued and the roller is lowered, it will meet metal across all of its face, with the metal being displaced sideways into the margins. The soft ferrite would react in a similar way to Figure A, but the hardened skin would resist that movement, and itself tend to move sideways. Rather like pushing two ends of a piece of paper together it will pucker and, depending upon the strength of the case hardening, will dip back below the mean plate level. When the top of this ripple has been removed, leaves a sharp and relatively deep indentation in the margin: a TRL.
It is these sharp edges, sometimes both sides of a TRL, that flags the difference between a TRL and a burr rub.
Inconsistent results?
So why don't we see TRLs on every second state? Because there were too many variables. Metal composition; efficacy of the case hardening process; the exactness of plates, rollers, and the setting process. A difference anywhere will dictate different results. Some plates were very hard even before they were softened. Other soft plates did not take the case hardening well (due to low carbon content?). The metal bulging out without going below the plate level, could have been removed without any evidence. Conversely, it is conceivable that the metal flowed into the plate body directly from the edge of the transfer roller.
Re-entry on hard or soft plates?
Long running discussion this one! The general view is that all re-entries were made on soft plates, which implies that plates went through the decarborisation process as practiced by Perkins Bacon, repaired and then re hardened. I don't buy it. Bringing the plate anywhere near cementing heat will risk ruining the plate, so it was not normally practiced. IMHO the one and only plate that was softened is 155. Bacon states:
“When this happened [wear], the worn impressions would be"re-entered"without softening the plate, and after this had been done the plate would often be left as it was: at other times it would be hardened a second time.” Vol 1, p.60
Interesting point is how many plates were re-hardened? Generally speaking, once an impression is re-entered, it tends to accelerate wear of that unit, no doubt due to the damage to the structure of hardened metal caused by the roller.
Another point in favour of re-entry onto hard plates is the fact that generally when a plate is repaired by re-entry, only a limited number of impressions were so treated. If the plate were soft, what would have stopped them repairing the whole lot? The other side of this coin is that any plate that was completely repaired was never hardened in the first place, viz 173-177. We are told that die 1, plate 1 was not hardened. Having looked through the excellent Sir Humphrey Cripps Nissen reconstruction on the Spink website I can confirm that none of the stamps from plate 1b exhibit a significant TRL, nor interestingly any from plate 5 although I did note a few examples from plate 10. Added to this unhardened list is of course 155.
What of TRLs that appear on an imprimatur sheet?
There are a few, and I was at a loss for a while. Then I found in Mr Brewer's article:
“An identical line appears on the left-hand margin of TE [of 2d blue plate 4] on all printings (see the Imprimatur sheet). This unit was either re-entered after the plate was originally hardened, making it a Fresh Entry, or it was subjected to excessive pressure from the Transfer Roller when first laid down. We think the former was the case.”
Then I remembered reading an extract from the Hardening book in the Perkins Archive re plate 82 of die 2 described “fire fierce, unequal, 4 heads injured.” These could well have needed repair, almost certainly by re-entry and would have been completed on the newly hardened plate, that is without softening. Perhaps a collector of this issue could shed more light on this point. Unfortunately, the Hardening books in the Perkins Archive do not start until 1863.
The change in width of transfer rollers.
It has been pointed out to me by an eminent philatelist that later plates exhibiting a TRL have the position closer to the stamp due to the rollers being slightly narrower than those used early on. I would argue that the width of the roller dictates where the TRL will occur, not if. The displaced metal is constrained by the roller until it reaches the rollers edge. Strictly speaking, it is the impression in relief that causes the displacement of metal that leads to the effect, not the edge of the roller.
Conclusions
I believe that the cause of TRLs has been mis-described. There is an added complication in that the definition of TRL and Burr Rub in common use is way too woolly, with these terms being interchangeable. They are not. Also, repair by re-entry was normally carried out on case-hardened plates, not in the the normally softened. This all leaves me in the uncomfortable position of having disagreed with some very big names. Again.
"Bloody typical."
The wife said.
AP
2013-18