The primary source book by Norman Collins, “The Crown Agents Requisition Books” - a compilation of postal supply orders from the Mandate era - was one of the last books I acquired, just a few months ago. Until then I’d built my philatelic library piece by piece, subject by subject, sparing no expense or effort in acquiring the books I needed to progressively understand different periods and subjects. Here though was a book whose name I came across from time to time but felt that as a compilation of orders it contained information that I wouldn’t really need know.


However as I became more immersed in philately and began specializing more in very specific subjects, like Mandate coil stamps, booklets and paper types, I found this source mentioned more and more, not just as a source of order quantities but as a primary - almost exclusive - source from which information was being extrapolated and conclusions inferred, sometimes by way of linkages between specific stamp types and orders which appeared in the compilation. Particularly when I began encountering stamps that didn't match the published data, such as dates of use or paper types used, I felt I had to get the book - urgently - and I did, but quite quickly I found many inconsistencies with the information and upon further research I reached the conclusion that this source is vastly incomplete and that at a minimum we must be very careful drawing conclusions from it. This article shares my research and findings:



Since the contemporary discovery of documents from the Mandate postal service in 1979-1980 by Emanuel Glassman (reference Marvin Siegel in the Israel Philatelist 1980 & 1981 vol. 31-32) and the subsequent assembly of the orders placed by the Crown Agents for the Mandatory Posts and Telegraphs department together paired with the recovered documents and carbon copies held by Arthur Hochheiser by Norman Collins, this compilation of orders published as the Crown Agents Requisition Books has become the cornerstone for first-hand research of the postal history of the Mandate.


By way of its simple and concise listing of orders by purchase order number, date, breakdown of items ordered plus additional comments, the compilation has lent itself to making easy and intuitive linkages between specific orders and their dates on the one hand, and philatelic material under observation on the other hand. Its influence is widespread: from the practical side it forms the basis for establishing dates of usage, quantities ordered and paper types used – and more influentially, catalogue classifications; and by extension the subsequent conclusions drawn by this assembly of raw information and derived conclusions has crystallized and defined our conception of Mandate postal history and the parameters of its scope, permeating virtually any book or article that has been published since.


The Bale and Dorfman Mandate catalogues are among the specialized literature that rely heavily on this source to draw inferences, publish production figures by stamp type, and establish their categorized delineations of paper type usages for stamps. Among the articles that have been published, David Dorfman for example links the existence of 7m and 8m Pictorials stamps on thin vertically ribbed paper to being part of a 1934 order (#6082) for 5m coils (reference “The Postage Stamps of Palestine 1918-1948 - 2nd Update” in Israel Philatelist issue 42 p.28, 1991); and Arthur Hochheiser (and others before and since) draws the conclusion that there was only 1 order of a 3m coil stamp (#2932) in 1928 (reference “The Palestine Paste-Up Coils: An Enigma” in Israel Philatelist issue 50 vol 2 p.8, 1999). These are a miniscule sampling of examples just for illustration.



Nevertheless, as illustrated above, I have an April 1933 postmarked 8m stamp on thin vertically ribbed paper which pre-dates Dorfman’s published first dates of use of July-Aug. 1934; I also have a 3m stamp on thin vertically ribbed paper postmarked Sept. 1932 – over 4 years after the only recorded order of this denomination on this paper. Clearly there is a problem – either with the empirical evidence or with the source information. And the empirical evidence, wider than shown here, is entirely genuine and legitimate.


Dr. Hochheiser, for whom I have tremendous respect as a researcher who always examined the critical technical-philatelic aspects of “how things work” and whose research work is both unique in its insight and groundbreaking in its conclusions, nevertheless made a disingenuous (and unfounded) comment in his article to help explain why only 1 order for a 3m coil stamp would exist: “The need for additional rolls was deemed unnecessary since this supplied the postage for printed matter only”. This probably unintended attempt to speak on behalf of unknown entities without documentary basis, in order to rationalize clearly seen but subconsciously accepted shortcomings in the records is a phenomenon that recurs in regards to conclusions drawn from the data in the Requisition Books. By contrast on this point, the philatelist Irwin Math wrote a piece extolling that 3m denomination as the ‘Workhorse of the Mandate’, summing up “No other stamp without change of color has been used for a greater period of time during the Palestine Mandate than the three mils of the Pictorial issue - twenty one years” (Israel Philatelist vol. 60 issue 5 p.22, 2009), and Math’s position on this point is more persuasive given the genuinely useful and ubiquitous use of the 3m denomination throughout the years of this stamp issue.


With Hochheiser’s comment as a segway the greater problem is, empirical evidence of stamps existing on different paper types than those listed in the Crown Agents Requisition Books, or existing before or beyond dates listed in that source call into question the completeness of this source. Furthermore, on closer study - even superficially - of the listings, a number of peculiarities immediately arise: we know from the documentation that Palestine’s stamps were produced in the UK and imported from abroad, but


  1. the record of orders often shows large time gaps between the stamp orders, sometimes as much as 6 months to a year or more; (ref: p.10/17/35/38)



  2. in specific areas such as the coil stamps – whose research is heavily reliant on this compilation – we see no orders for coil stamps between 1930 and 1933 and again between 1936 and 1938, a period of tremendous growth and development in the postal services domestically and abroad; (ref: p.27)


  3. in another specific case the high denomination Pictorials (250m/500m/£1P) were only ever ordered two times, in Sept. and Oct. 1941 – and then never again (p.22); from data I will show below, 158,000 parcels were mailed that year, with an average of 170,000 a year up to 1947 (peaking that year at 230,000) – no additional high denomination stamps were needed?
  4. as regards a seeming inconsistency in the regularity of orders, the 200m Pictorials for example are recorded being ordered once in June 1933 and again six years later in June 1939 (p.19/21);
  5. No consular fees stamps were ordered for two years, between 1923 and 1925 (p.39);
  6. Almost no orders are recorded for 1948, although the post office continued to function apace until its wind-down in mid-late April.

Indeed for a record which should be recording the purchase history of a postal service - considered the best in the Middle East - for 30 years, the 40 or so pages of actual orders (some of which are filled up by images of documents, leaving the net space allotted for the orders to less than 40 pages) is thin to say the least. Obvious postal articles are missing from its scope: the actual vending machines for the coil stamps, the pillar boxes for the submission of mail, postmarks, any office supplies and equipment required by the network of post offices in Palestine, vehicles for the transport of mail, and any of the vast network of telecommunications equipment that was part of the Posts and Telegraphs’ domain. I understand that certain subjects in total may not appear in the compilation but I mention these omissions to draw the readers’ attention to the fact that much material is missing here - including in the areas that the compilation does cover. If I sharpen this point just a little more, I should mention that in [British] stamp production, one source provided the stamp sheet paper, another source printed the stamp, and another source watermarked the paper – three contracted sources of which only one, sometimes, is listed here; all required purchase orders from the Crown Agents.


Here we stumble upon a delicately presented - but overt - oversight, perfectly capturing my concerns about the Requisition Books, and this will serve as the jumping board for my research:


a glance at the section on Mandate postage stamp booklets reveals that only 4 orders are recorded, one for each year of 1929-1932, representing what the catalogues presently classify as Dorfman’s 1st series and Bale’s 1st and 2nd series of booklet stamps - but omitting the subsequent issues that are known to have been produced until at least 1939 (i.e. series 3 and 4). Addressing this absence Collins notes dryly that “other booklets are known and listed in specialized catalogues”.


Delicately deflecting potential concerns for the omissions - as Hochheiser presumed to speak on behalf of the involved authorities - so too Collins concludes authoritatively “There were only four printings of stamp booklets in England as will be seen from the Crown Agents’ Requisition Book entries… All booklets other than those sent out to Palestine by the Crown Agents, must have been made up locally in Palestine”. This is a clever but disingenuous formulation: without documentary basis Collins is stating conclusively that a) only these 4 orders were sent to Palestine; b) other booklets albeit are known - but these weren’t sent to Palestine, they were produced there and therefore these would naturally not be recorded in the Requisition Books. Collins sees the deficiency in the record but subconsciously rationalizes it. In his opening remarks on this compilation he alludes to information only being as complete as the surviving records, but he doesn’t leverage this awareness to confront the information and test if it’s reliable enough to draw conclusions from it.


His statement implies that there is in fact no omission in the records at all and that the subsequent issues (presently catalogued as issues #2-4) “must” not have been imported in any specially prepared form but rather cut out from scratch, presumably from standard shipments of stamp sheets. That deft summarization – and all that it represents in allaying fears that the compilation may be incomplete – has worked for the last 40 years and amplified the far-reaching conclusions drawn from this record… except for one tiny oversight which undercuts both the point in question and so the completeness of this record:


Collins, Dorfman and Bale illustrate both stitch-bound and staple-bound booklets as having been part of these documented orders of 1929-1932 – but British-made stamp booklets of this period (1917-1976) were all stitch-bound except one staple-bound booklet issue produced in 1969 (Scott #BK126); the British also produced stamp booklets throughout the Second World War even though Dorfman writes that due to paper shortages booklet production ceased with WWII – how does he know that? A search through the press archives (search the National Library in English & Hebrew: reveals that stamp booklets of various agencies and bodies existed in Palestine throughout the War


As such the Mandate booklets bound by staples – even these early ones – were domestically assembled from shipped guillotined panels of stamps and would therefore be “locally made” like all the others that are unlisted in the Requisition Books compilation. Indeed the Palestine Post of 4 Aug. 1949 (p.2) credits a local company “Near East Advertising Company” and its owner Gabriel Roos (who won the franchise to produce Israel’s first stamp booklets) with being the driving force behind the Mandate’s stamp booklet initiative. A difference in using stitching versus staples is a change in specifications, not a trifling detail. And if we relying on the habit of the published record to document for instance that booklet orders included both guillotined booklets/panes as well as excess sheet columns to be used as regular postage, if booklets really were produced from scratch in Palestine we should expect to see a breakdown of a standard stamp-sheet order indicating how many sheets are being earmarked to the production of stamp booklets and how many for regular use. Such information is not in the complied record.


The absence of the orders of the known subsequent stamp booklets from these records shows clearly that the record is incomplete, and if this record is incomplete we cannot make far-reaching conclusions based on the [few] orders that are recorded. 


And how many orders are missing? A momentary aside:

Initially I thought it may be the number of purchase order numbers that are missing in the sequence of those that are listed. I mention this briefly as a lesson for the reader, because it’s incorrect but instructive: the error in my assumption was that the purchase order numbers were not issued by the Palestine Government or its departments but by the Crown Agents themselves. As such these numbers as they appear in the compilation are random and entirely the product of the CA and their numbering system.


To understand the various document numbers observed in the pipeline of correspondences for purchase orders, in broad strokes, each correspondence sent by a government department or the Crown Agents or a solicited business was issued its own serial/filing number (number “1a” in image): this was written and date-stamped by the sender; the receiving office would likewise date-stamp the received letter and issue it its own departmental filing number (number “1b” in image). Any future references to this specific letter would reference it by its assigned filing number (and those letters referencing it would themselves be date-stamped by both the issuing and receiving offices, each one assigning that subsequent letter its own filing number – behold, bureaucracy) – see number “2” in image.


Here then when the Mandatory government (or a department thereof, such as Posts and Telegraphs) approached the Crown Agents with a request for the purchase of material, they would refer to this request as an “Indent” – ‘an official request for goods’, in British parlance (the word is both a noun and a verb) – and give this Indent a number (number “3” in image). Subsequent correspondences between the Mandate Government and the CA on a certain “Indent” would reference its number (all the while generating new filing numbers between each office for the letter exchanged on the matter).


The purchase order number itself was generated by the Crown Agents as “Palestine [purchase order number]” from the moment the CA approached businesses for tenders (see number “4”): in other words this number existed even before a tender was accepted and a contract signed. Both tenders by the companies and the purchase orders issued by the CA reference the purchase order number, the indent number (issued by the Mandate Government) and the relevant Department in the Government, here “P & T” (Posts and Telegraph).



That closes the cautionary tale of mis-interpreting the purchase order number sequence, but... many orders then are missing from the compilation?

Here I’m aided by the first-hand data assembled by Daniel Rosenne in his book “Communication in Eretz Israel During the British Mandate” (2019 ed., published by the Association for the Commemoration of the Fallen Soldiers of the IDF Signals Corps): he mercifully culled the department’s annual reports and assembled a table of statistics showing the annual processing of mail from 1920 to 1947 (although for certain years and mail types the data is incomplete). I display it here for the benefit of the reader who may not have seen these statistics before (p.287-289):



I compared the postal service’s annual data on mail processed to the quantity of stamps and stamped postal stationary ordered (i.e. produced) each year, as shown in the Requisition Books, converting sheets and rolls into their equivalent number of stamps (based on the denominations involved, some being 200 stamps per sheet and others 250, or 500 in a roll etc.):


Although Rosenne tabulated data until 1947, I stopped my tallies at 1937 because I began seeing that the total amount of annual stamp production was consistently not keeping pace with the quantity of mail being processed annually.


What I found – as we see – is that the amount of mail processed each year outpaced the quantity of stamps being produced that year. A rational assumption would be, allowing for stampless (i.e. “Official”) and metered mail to be considered statistically insignificant, that there should be at a minimum parity between the quantity of mail sent and the number of stamps and stamped postal stationary produced – that there should be enough stamps to frank the mail sent at a basic rule-of-thumb rate of at least 1 stamp for every piece of mail sent. Here we see, other than oddly wild variations in stamp production from year to year, that on an almost annual basis, stamp production was usually 40-50% below the amount of mail sent. Over 17 years I tallied – even with clearly incomplete mail processing data, like for 1921 – we find that if we rely on the compilation of orders, the Mandate postal service was 110 million stamps short of supplying franking to the mail it was sending. Clearly something here is amiss.


NOW: is this possible? Is this the norm? I tried to compare the Mandate to her mother, the UK and even to the US, expecting that detailed statistics would be forthcoming. For the US this information was much easier to find, whereas for the UK I was able to establish a link between mail processing and stamp production for just 1 single year (1939-1940).


For the purpose of drawing conclusions let’s look at the tabulation of similar annual US stamp production versus annual mail processing – here for certain years the data is aided by an extrapolated value representing metered mail, and this indeed helps complete the picture for the US:


Here, with almost perfectly complete information we see that indeed it was possible for stamp production to be substantially less than the quantity of mail being processed in a given year - usually 25% less on a consistent basis, up from 15-20% in earlier years - but that shortfall was almost perfectly made up by the quantity of [stampless] metered mail sent through the posts (as per my comments below the table, I extrapolated metered mail's "percent of revenue" as being like "percent of stamps produced" implying one-for-one parity between the number of metered mails and their face value - for 1937 and 1939 that worked; 1932 is anomalous as the total calculated production of stamps is 90%, where the metered mail is 14.87% of that total, but this merely means that the nominal value of that metered mail was low so that 14.87% "share of revenue" is not equal to the missing "share of production" of 24.87%. In all likelihood metered mail really was almost 25% of the mail processed but ony accounted for almost 15% of the stamp revenue that year).


By contrast, Palestine’s stamp production is bombastic in its fluctuations, and although we don’t have a statistic for the amount of metered mail sent we do know (from Sacher in “The Postal Markings of the Palestine Mandate”, 1995 p.86/170/216-217/239-249) that it was small, only being used from 1930 (until 1948): 25 machines in Jerusalem, 19 in Haifa, 4 in Jaffa and roughly 40 in Tel Aviv. Just from mail that I myself examine I doubt that 5% - one out of 20 pieces of mail - of what I see is metered mail (and even that may be an exaggeration): at best, that still leaves roughly 23% of the franking for mail on aggregate for 1921-1937 unaccounted for - and in some years much more.


From this alone we know that almost 25% of the postage used for this period is missing from the Requisition Books compilation.


And if we throw the proverbial “spanner in the works” by referring for a moment to the single British statistic that I found, for 1939-1940 we find that 7,360,000,000 items were processed by the British post (source: Royal Mail statistics compiled by, and 8,395,000,000 stamps produced that year (source: “The British Post Office - A history” by Howard Robinson, 1948 p.444: "by 1939 the Post Office was finding it necessary to print an average of 23,000,000 stamps every day"). That comes to a 114% relation of stamp production to mail processed that year (14% more stamps produced than mail processed) – far in excess of the US and not yet accounting for the effect of the more widely used meter mail in the UK.


I should add, as a curiosity, that when I first assembled the US data I came cross production figures mentioned in terms of “sheets” only and extrapolated their number of stamps on the basis of 200 stamps per sheet as I had observed (Scotts Catalogue in their own introduction to US stamps, 2006 ed. P.21A/22A write that actually 400 subject sheets were used for the majority of regular postal issues, and references 200 subjects per sheet for some of the commemoratives): had my extrapolation been borne out by the subsequent stamp production figures I found, US stamp production relative to mail processing would have been anywhere between 115% and 124%. Nevertheless and quite oddly the documents which provided the actual stamp production figures yield an average of about 105 stamps per “sheet” – as if the word “sheet” actually means “pane” from a standard US stamp plate (sheet). [August 2021 inline postscript: it had occured to me that my calculated results for US stamp production versus mail processing in the table above is a little "too exact" (i.e. for 1937 and 1939) - it would appear that there was no excess stamp production, no margin of error in production, for collectors, dealers, natural overproduction etc. I would submit that perhaps my extrapolated measure of metered mail is too stringent and that perhaps the value of the metered mail may be less than the quantity of mail sent (I was using parity of 1:1 as explained above; perhaps 25% share of revenue is really, say, 30% of the mail processed). It may also be that, as I wrote just above here, there is something odd about how "stamp sheet" and "stamp quantity" as provided in the sources cited yields such a low stamp count per sheet - a small margin of error in that data multiplied against a few hundred million stamp sheets may be enough to generate the "excess stamp production" that seems to be missing from my table above.]


As we can see stamp production data – orders – are indeed missing from the Requisition Books. If we try to quantify how much is missing I would take from the 1921-1937 data an average of the production to mail processing for the years which do not have an extreme result of 150% production or more, and that would leave us with a relation of 64% stamps produced to mail processed. Allowing for as much as 6% of the volume to be accounted for as stampless (Official mail and metered mail), this would bring us to 70% - and then at a minimum, without allowing for any excess production, 30% would still be unaccounted for, for the mail that was sent: 119.5 million stamps. In other words at least 41% of the orders are missing from the Requisition Books - just for the period 1921-1937 alone.


Parting Shots

All this was in aid of testing the completeness of orders for postage stamps. But what about revenue stamps? For the US in 1939, almost as many revenue stamps as “ordinary postage stamps” were produced – 14,684,696,000 to 15,073,796,000 (reference p.586 of “Treasury Department Appropriation Bill 1940”, 1939); for Palestine in 1937 - the last year of my data examination - the requisition books record… none - no duty stamps were needed to generate income for the Mandate. For 1936 we see 4.2 million revenue stamps in the documented orders (as 13% the quantity of ordinary stamps produced that year). Those revenue stamps represented a nominal value of £7,127… as against the year’s liquor license fees, alcohol excise fees, business registration fees, patent fees and trademark fees totaling £7,497… plus broadcast license fees of £10,473 and any applicable fees paid with revenue stamps from the year’s excise duties totaling £359,498 (source: “United Kingdom Report on the Administration of Palestine and Trans-Jordan for the year 1936”, 1937). Again, the compiled stamp orders seem to be coming up short.


A late edit (Nov. 2021): For the sake of an interesting comparison, I found the following information in the Palestine Post of 6 April 1939, reporting that the Kofer HaYishuv initiative ("Jewish Redemption Fund") of the Jews in Palestine - indirect contributions in the form of voluntary taxes on entertainment, cigarettes, transport, liquors and other various commodities) - raised £58,655 on 17,699,000 revenue stamps issued, for the 6 months ending in Febuary 1939, and that each Jewish inhabitant in Palestine contributed an average of 130 mils. As the Requisition books list an order for revenue stamps on 25 Feb. 1938 and another for 14 July 1939, if we consider the value of the order from 1938 it comes to 1,380,400 stamps representing a nominal value of £201,190 - assuming (riskily) that this was the entire production of revenue stamps for the year. Kofer HaYishuv did not replace the Mandate's revenue charges, it was an additional taxation to them (for the Jewish community, the "Yishuv"). Here we see that as a community numbering about a third of Palestine's population, in the space of half a year on a smaller scale of taxable activities, it generated about 30% of the nominal value of the revenue stamp order for 1938 intended for the country in total for [almost] the full year.


There are many other critiques and observations that I have about the Requisition Books compilation but I think with the information I’ve shared above I’ve made my point. This is an incomplete primary source from which we have to be careful in inferring conclusions, and as a general proposition we do have to release ourselves from preconceptions formed from its information and go back-to-the-basics and do our own homework researching issues from scratch, from the bottom on up.