File No.: Case File # 19
Title: "The Lost Aerial Photos of Arnhem"
Subject:  Allied Aerial reconnaissance photographs showing Nazi armored units in the Arnhem area prior to Operation "Market Garden".
Investigation made at: Deelerwoud Forest, municipality of Ede and Wageningen University, Province of Gelderland, The Netherlands.
GPS Location: 52°04'45.4"N 5°53'54.3"E
Period Covered: 12SEP1944-22FEB1945
Date: 27NOV2018
Case Classification: Imagery Intelligence (IMINT / PHOTINT), Aerial photography analysis and
interpretation, Evidence provided for error in Royal Air Force Air Historic Branch publication.
Status of Case: Unsolved

The 1977 Hollywood production A Bridge Too Far features the remarkable story of a British Intelligence officer in the Airborne Corps (named Brian Urquhart in reality but called “Major Fuller” in the film to avoid confusion with Airborne General Roy Urquhart played by actor Sean Connery), who requests a last minute photo reconnaissance sortie and gets the approval for it. The result of the Major’s determination however, an oblique photo showing German tanks and halftracks concealed in a countryside tree line, is not appreciated by his superior, General Frederick “Boy” Browning, commanding the 1st Allied Airborne Army.
The photo shown in the film (which is clearly fabricated for the production of A Bridge Too Far as it shows the photo which the pilot in a previous scene about the photo reconnaissance sortie took) has intrigued this agency for decades and raised the desire to learn more about it.

(click to enlarge)

The Hollywood aerial photograph

Several documentaries have shared our interest in photo intelligence indicating enemy armored concentrations in the area of operations of "Market Garden".
For example: "Battlefield Detectives"; a forensic documentary television series that aired on the History Channel from 2003 to 2006. In Season 2 "World War II: Operation Market Garden" of 12NOV2004 we hear Peter Caddick-Adams of the Royal Military College of Science in the United Kingdom say "In the week prior to the Arnhem operation being launched on the 17th of September, there is a suggestion that German armour is spotted by aerial photography in the Arnhem area". The narrator continues: "But no one since has seen these photographs. Peter believes that if they can be found, and German tanks can clearly be seen, it would be compelling evidence that Operation "Market Garden" was flawed from the outset. The search leads him to the archives at Keele University. But it's frustrating. Although a staggering eight million images survive from the Second World War, at the end of the conflict, the cataloguing system was simply thrown away. Finding images of a particular operation or place is like looking for a needle in a haystack." The documentary then  shows an example of an areal photo of Arnhem bridge taken prior to the operation and showing traffic followed by Caddick-Adams talking about details such as tanks and trenches which can be seen in such photographs.
Narrator: "It's these traces of German activity that Peter is searching for in the images from Arnhem. [...] But it's the missing images from Arnhem that Peter really needs."
In answer to the question of Caddick-Adams if he has found the equivalent for Arnhem, the projector operator in the Keele Archives says: "Unfortunately in the material that we've so far catalogued we have not found anything. it's not to say that we don't have anything but maybe in years to come we will find the photograph that you're after.
New Evidence
But in 2015 it seemed that Dr. Sebastian Ritchie of the Royal Air Force's Historical Branch had at last found the "pièce de résistance"
of Allied intelligence for Operation "Market Garden".

In "Arnhem, The Air Reconnaissance Story", (Crown Copyright MOD 2016, published (online
until shortly after this Case File was published) by the RAF's Air Historical Branch) the following photo is presented with the accompanying contemporary analysis and interpretation diagrams:

(click to enlarge)

On pages 15 & 17 of the RAF AHB study we read:
"The Spitfire routed east, north of Arnhem and across the main Arnhem-Apeldoorn road, before banking and commencing a westward run just south of the village of Loenen, orientated slightly north of the Luftwaffe airfield at Deelen.
Seconds later, Fuge was flying over woodland known as the Deelerwoud, northeast of the airfield.
His first frame was numbered 4001; [footnote 67] frame 4015, his fifteenth (out of a mission total of 942 frames) differed from every other in so far as it contained visible markings and lettering,
indicating that it was the subject of a detailed report or briefing.
Yet, while the area concerned was immediately next to a very large military target, the lettering clearly did not relate to the airbase in any way, shape or form.
Sadly, the report that accompanied frame 4015 could not be traced, but a high-resolution download of the photograph ultimately explained  lettering, albeit via enlargement and some digital enhancement.
The interpreter’s task was probably to check each area of the image for indications of enemy activity. In Areas A, B, D, E and F, he had only to confirm its absence: there was nothing to report.
The contrast with Area C could hardly have been more pronounced.
Here, multiple German armoured fighting vehicles (AFVs) and other vehicles can be observed near the intersection of two woodland tracks, apparently halted while moving south. Some are partly obscured by tree cover, while others are in the open; camouflage measures are not in evidence. While there is insufficient resolution and too much cover from trees or shadow to provide more than a few reliable identifications, the larger tanks include Panzer IVs of early design, with short-barrelled 75mm guns; there are also smaller tanks, including Panzer IIIs, which are, again, early models equipped with 37mm guns. Some of the tanks have rotated turrets, probably to create space for maintenance work or fuelling – a routine procedure.

Footnote 67 reads as follows: 67.! 1&bron=BWUR_WOII_RAF_USAAF&sortienummer= 106G%2F2816 (accessed on 2 July 2015).

Dr. Ritchie's findings called for further investigation into the case of the “Lost Aerial Photos of Arnhem".

Observation of the T-junction in Deelerwoud Forest
On Sunday 18NOV2018 Battle Detectives proceeded to the areas indicated in Ritchie’s analysis as
“Probable supply dumps” and “Tanks, AFVs and other vehicles” on frame 4015. We found these locations by comparing the terrain features in frame 4015 with satellite images on Google Maps and pinpointed the GPS-coordinates. These areas are located along an unpaved track named Hoge Delenseweg in a wooded area named Deelerwoud (Forest of Deelen) which can only be accessed on foot. We found the T-junction with the supposed vehicles and a gate at the end of the track close to where the supply dump would have been. Other than a significant number of large rectangular shaped pits, placed diagonally in the verges of the dirt track, we found no man-made structures in close proximity of the forest path. For some time it was theorized that these holes had been constructed to park AFV's in during bivouac.

(click to enlarge)

Fig. 1  The current position of the T-junction in Deelerwoud.

The T-Junction on Hoge Delenseweg,
the gate at the north end of the track looking south,
and the airspace above area "C"

The partially overgrown man-made dug-outs along Hoge Delenseweg in the Deelerwoud Forest

An enquiry made via e-mail on 19NOV2018 with the proprietor of the forest, the "Vereniging Natuurmonumenten", revealed that "the way you have described, these are waterholes. These are drainage spots to allow rain water to run from the road top. This has happened often in the past and these are maintained as well."
Discoveries made at Wageningen University
The discoveries of this agency made on Wednesday 21NOV2018 in the Special Collections Library of Wageningen University & Research are amazing and prove Dr. Sebastian Ritchie’s thesis wrong.
Apparently Ritchie has only accessed the photos he used, online and ordered a high-resolution copy of frame 4015 which was the only one with markings on them.
It is our theory that Ritchie has been looking for photographs taken of the area around Arnhem taken during photo reconnaissance missions shortly before Operation Market Garden.
He thus found the series taken on 12SEP1944 and zoomed in on the photo with the markings on it.
His theory is that the letters A, B, D, and F indicated areas with no enemy activity and C did have enemy activity. Ritchie then sees tanks in such detail that he can make out older Panzer III types and by deduction the Hermann Goering Parachute Panzer Training and replacement Regiment which saw action on 17SEP1944 in Son and later in Nijmegen.

(click to enlarge)

Fig. 2 Now & Then comparison of the location where one of the
PzKpfw III of II./FJR Ersatz und Ausbildungs Regiment Hermann Göring
was knocked on 17SEP1944 in Son in the book "Operation Market Garden"
(ISBN-13: 978-1-61200-586-7) by Simon Forty and Battle Detective Tom

Backside of photo
However, in the Wageningen University & Research collection of World War 2 Royal Air Force photos we took a look at the backside of the print of frame 4015 and found that these markings were made for the 1949 Dutch publication of the “Boor & Spade” (Drill and Spade) periodical about soil and ground mapping.
Frame 4015 is used as photo 18 in this issue and shows (as can also be seen in the English summary: "High sandy soils on the Veluwe to the north of Arnhem. A road. F young arable land. C wood. D shifting sand with heath.
B and E differences in growth in the wood owing to gravel-pans and loam-pans.

(click to enlarge)

Special Collections on Wageningen University Campus
and the backside of photo 4015

The Original Dutch caption for section "C" is "High sandy land on stowed pre-glacial" but not: "enemy activity".
The markings have been added in 1949, not in 1944 and for the purpose of illustrating various types of soil in the Veluwe area in "Boor & Spade".

 (click to enlarge)
"Don't judge a book by its cover"
The 1949 issue of "Boor & Spade" and the use of frame 4015
as an illustration of various types of soil in the Arnhem area

We also found that by looking at both frames 4014 and 4015 through our stereoscope, using +1 reading glasses
and a strong pocket lamp shining on it, we could create a 3D-image of the area.
We were able to make out objects on the T-junction which we visited on 18NOV2018, but found it impossible
to tell if we were looking at either wooden crates or vehicles.
We did observe, however, that the objects in the photo taken on 12 (and on 19; more later) SEP1944 are not in place at present.

Imagery Interpretation
On page 40 of Dr. Ritchie's "Arnhem, The Air Reconnaissance Story" we read:
"It was a high-altitude sortie that located armour north of Arnhem on 12 September 1944, including Panzer IIIs
and IVs ‘tucked in underneath woods’. However, they belonged not to II SS Panzer Corps, but to the Hermann Goering Parachute Panzer Training and Replacement Regiment, a formation that had long been using the same
area for training, as the Allies well knew. A key component within this narrative was the order issued to the
Hermann Goering Regiment on the previous day, dispatching them to Eindhoven. When they were caught on
camera, they were arming, refuelling and moving south in preparation for this deployment. It was probably reasonable to identify the tanks as a potential threat to 1st Airborne Division but the issue was not straightforward. The appearance of modern tanks such as  Panthers or Tigers in the imagery would certainly have provided grounds for serious concern, but the prevalence of older model Panzer IIIs and IVs could well have suggested to an experienced intelligence officer that they belonged to a second-line unit of questionable combat capability.
A reasonable conclusion might have been that the photograph reinforced the broader intelligence picture of
German militarisation in the Market Garden area but did not necessarily point to a specific threat from a first-line
panzer formation at Arnhem.

Dr. Ritchie suggests here that the type of AFV can be discerned from the photo and that from it, the unit can be deduced and that these vehicles were moments away from deployment elsewhere.
The Wageningen University librarian helped us and found that another photo reconnaissance sortie was flown on 19SEP1944; two days into Operation “Market Garden”.
Sadly just the prints of two frames, numbers 4201 and 4202 in the series of this sortie, which show the T-junction in the Deelerwoud Forrest, are missing.
The librarian did show us the high resolution scan of these frames on her computer screen. In our perception these images showed the same objects along the T-junction and the "Supply Dump objects" a bit further north sitting
on the dirt track, as can be seen on the photos of 12SEP1944. If these were working AFV’s they would have been deployed.
Therefore, if these were the Panzer III’s of the Parachute Panzer Training and Replacement Regiment Hermann Göring they would not have been at the T-junction on 19SEP1944.
In the following comparison we see the image used in Dr. Ritchie's hypothesis about the objects seen on
12SEP1944 being tanks that were deployed south of this location on 17SEP1944, at left. At right we see the same location on 19SEP1944; two days into Operation "Market Garden". Both images show the high-resolution digital
scans of the photos on the librarian’s computer screen and enlarged to the biggest possible size without the pixilation rendering it unsuitable.

 (click to enlarge)

Fig. 3 comparison of the T-junction on 12 and 19SEP1944 from the digital scans

There are also photos taken on the 22FEB1945 and on these, the supply dump objects are now not on the track anymore but next to it in the woods. It seems that the objects next to the T-junction are gone on that day.
Therefore it can be concluded that Dr. Ritchie was drawn to Section C in the aerial photo with the objects because
of the markings on the photo print of which he erroneously assumed that these were made by a photo intelligence analyst.
He saw armored fighting vehicles (AFV's) in some shapes on the photo of which he assumed that these AFV's belonged to the Parachute Panzer Training and Replacement Regiment Hermann Göring.
This unit saw action on the 17th but the aerial photos of the 19th show the same objects still in place.
It is therefore highly unlikely that photo 4015 shows German armored fighting vehicles which were deployed during Operation Market Garden.
Validation the hypothesis
From our findings in Wageningen University and in Deelerwoud we have developed the hypothesis: "Frame 4015 of RAF sortie 106G/2816 of 12SEP1944 and frame 4201 of RAF sortie 106G/2996 of 19SEP1944 both show the same objects on the T-junction in the Deelerwoud and are most likely not German tactical vehicles as theorized by the Royal Air Force Air Historical Branch (RAF AHB)".
To validate this hypothesis it is imperative to test its falsifiability.
Below are five statements and their argumentations as contrary views on our hypothesis, which are in turn answered by our validating observations.
Falsifying statement 1: The book where the lettering was found is not available in the UK.
A further difficulty arose because most of the interpretation reports that went with the photographs have been destroyed. The lettering was discussed with several members of the Medmenham Association, which specialises in the history and heritage of air reconnaissance, and they all assumed that it was written on during the analysis/interpretation of the photo in September 1944. It will obviously be necessary to amend the Arnhem air reconnaissance study to take account of the discovery in the Special Collections at Wageningen University.
Validating observation
This statement confirms that an assumption was made about the lettering on the print of frame 4015 which started the focus in this photo and subsequently the rest of the RAF AHB's theory which was solely based on a digital scan.

Falsifying statement 2: Frame 4015 of 12SEP1944 wasn't the only instance in which the objects on the T-junction in Deelerwoud were photographed and recognized prior to the operation.
An RAF Mosquito also photographed our area of interest on 12 SEP1944. In this instance, the mapping created by the Allied Central Interpretation Unit has survived in the UK, and an 'A' has been marked on frame 4023 at the precise location where the tanks were photographed. This letter would have been matched to another 'A' written onto the photograph itselx`1f (or a blow-up of part of the photograph), and stapled to a single page interpretation report. As an example of this procedure, see the photo of the V2 launcher on page 14.
Therefore, we can be 100% certain that this specific place was of significant interest to the Allies. In addition (on page 22), it is recorded in the UK National Archives that a Form White (a fast, high-priority interpretation report) was generated following the Spitfire sortie on 12SEP1944, and that the Deelen area was the subject of this report.

Validating observation
On the illustration on page 24 we see a section of map G.S.G.S. 2541 N. of Arnhem (Holland). The map supposedly shows what the cameras in the Mosquito sortie have photographed and on it several rectangles have been drawn. One of these rectangles has the number 4023 written in it and it covers the Deelerwoud. There is no other marking inside this rectangle; let alone a letter ‘A’. On page 23 we read however: "A plot supplied by the National Collection of Aerial Photographs records the track flown over Deelen by the Mosquito, and shows that its cameras were activated directly over the woods where the German armored unit had previously been spotted. The plotters at the Allied Central Interpretation Unit, RAF Medmenham, had also marked this area with an ‘A’". An image of the marking of this area with an 'A' by the Medmenham plotters was NOT provided. This only proves that another sortie was flown over the Deelerwoud on 12SEP1944. It is very likely that this plot and said 'A' marking is not available today. The photograph on page 14 is merely an illustration of what a letter 'A' written on an aerial photograph looks like. The hypothesis "There is supposedly a letter 'A' on a photo taken in the area of the Deelen Luftwaffe base, hence it must refer to objects on a dirt track north of this base" isn't a logical one.
There is no doubt at all that the Deelen area was an area of great interest to the Allies as this was the location of Fliegerhorst Deelen; a large German Luftwaffe air force base which could pose a great threat to Operation "Market Garden" in the form of enemy fighter planes and anti-aircraft artillery.
Falsifying statement 3: Two highly qualified RAF intelligence analysts provided their input on the theory about German tanks.
Their view was that the 12SEP1944 and the 19SEP1944 photos only appear similar at low resolution. At high resolution, it is clear that the 12SEP1944 photo is far more three-dimensional, showing objects on the ground that project up towards the camera. These are rectangular, suggesting vehicles, and the sun reflection off at least four turrets is the classic indicator of armoured vehicles. They appear as 'pin-heads'.
Validating observation
We have looked at both prints of the frames taken of the T-junction in Deelerwoud (frames 4014 and 4015) in person. We used a stereoscope and could not discern detailed features on the square shaped objects sitting along the T-junction; let alone identify Panzer III's, their 75 or 37 mm guns in turrets pointing in various directions, or self-propelled guns and halftracks. Our method was the (3rd) next best thing after (1st) having either irrefutable evidence of presence of these vehicles there and then, or (2nd) access to the negatives of the photographs. For "Arnhem" the RAF AHB made use of digital scans ordered from the prints kept at Wageningen through the Dotkadata company which is the next (4th) optional method in a descending scale of suitable means of analysis.
Ordering of the RAF photos is through this website and it lists the highest resolution as: gem. (average) 9500x11000 pixels.

(click to enlarge)
Fig. 4  What an object known to be a Panzerkampfwagen III looks like
in a close up photo and on an aerial reconnaissance photograph
(frame 3113 from sortie 106G/3010; NCAP-000-000-039-722,
Wolfswinkel; North Brabant; The Netherlands
 Copyright HES,

The diagram in Fig. 4 shows proof of the presence of a Panzerkampfwagen III in a meadow alongside the road to Arnhem at Wolfswinkel North of Son (first in the scale of evidential value).
The AFV sat there from 17 SEP1944 onwards as it was destroyed by Allied fighter planes on that day.
The insets are sections of a digital scan of an aerial reconnaissance photo taken on 19SEP1944 showing the meadow with the AFV in it (fourth on the evidence scale).
The vehicle appears as a white rectangle and without the knowledge from the close up we might as well be looking at a discarded refrigerator.

Falsifying statement 4: The German vehicles seen on 12SEP1944 are gone in the 19SEP1944 photo.
By contrast, the 19SEP1944 photo is one-dimensional: there are no vehicles there. You are looking at the ground. When armoured vehicles – especially tracked vehicles – park anywhere, they scar the ground, turning it into bare earth or mud. The parking areas in this photo had probably been used many times over by vehicles using the supply dumps (which are visible on both dates), so the ground would have been heavily scarred. These scarred areas are completely different in appearance from the equivalent – occupied – areas in the 12SEP1944 photo (see for example page 20).
Validating observation
Look at the comparison in Fig. 3. The difference between both images is that the 19SEP1944 photo is less focused but not that it would lack a dimension. The falsification theory has it that 19SEP1944 photo shows the traces of (the tracks on) the vehicles shown on the 12SEP1944 photo.
We have been in Deelerwoud in November 2018. We observed that the T-junction is in a narrow dirt road and that the soil in the woods is sandy and dark of color when wet or moist. This was the weather at the Royal Netherlands Meteorological Institute at De Bilt (approximately 50 kilometers from Deelerwoud) in September 1944:

(click to enlarge)

These data paint a picture of a rather rainy start of the month with chilly temperatures causing a moist if not wet forest floor with a dry top layer. The latter is why the dirt track in the Deelerwoud stands out in the woods like a bright line. The difference in color of the objects along this track, between the photos taken on 12 and 19SEP1944 is minimal at best and not such that it warrants the assumption that we are looking at vehicles first and at their marks in the ground a week later. Vehicle tracks in soil this moist, would leave dark trails. Moreover, the (new) theory ("the rectangular shapes in the 19SEP1944 photo are the imprints in the soil of tanks on those same spots as in the 12SEP1944 photo") is highly implausible. A tracked fighting vehicle that drove away from its parking spot does NOT leave behind a contour of its shape on the ground; unless it had sunken hull-deep into the ground. When tracked vehicles move on unpaved soft terrain, they tend to leave behind trails that show up as a double line on aerial photos, like this:

American tanks of the 2nd Armored Division move through soft terrain in Germany, 1945
(C) George Silk—The LIFE Picture Collection

It is therefore that we consider it most likely that in both the 12 and 19 SEP1944 photos we are looking at the same objects. It is a misfortune that the prints of the images taken on 19SEP1944 are missing from the Special Collections in Wageningen University as a stereoscope comparison might reveal more knowledge on what is or isn't to be seen on the images.
Falsifying statement 5: The enhanced photo of 12SEP1944 shows objects that can be recognized as tanks.
In the detailed blow-up on page 38 of the RAF AHB publication, one can clearly see the turrets of both tanks, the frontal armor and cupola of the Panzer IV, and the gun and rear hatch hinges of the Panzer III. The two RAF analysts were 'very happy' with these identifications and with the 'certain’ ,'probable' and ‘possible' labels that were used. One cannot deny that these are tanks of the type identified by Brian Urquhart (and Tony Hibbert); tanks of exactly the type that later showed up at Son. In addition, there is no mention of the fact that the 12 SEP1944 photo shows tactical objects moving along the road, both in the wood and slightly to the north (see page 21). Again, these objects are entirely absent from the 19SEP1944 imagery.
Validating observation
When attaching so much value on the input of these analysts it is important to determine when the moment was that they were happy with the labels ‘certain’, ‘probable’ and ‘possible’. How were they asked to comment on these labels? Was this after the labels had been added to the image or were they asked to look at the junction without any prior knowledge?
On page 38 are two enlarged sections of frame 4015 captioned “Digital enlargement and enhancement of Frame 4015, 106G/2816, 12 september 1944”. This is the illustration on the right, undoubtedly the 'enhanced' image:

(click to enlarge)

Here, the digital image has been enlarged beyond the level of the pixels in the resolution as provided (9500x11000 pixels) and the images starts showing horizontal and vertical lines that are newly added by the software. Look for a reference at the tree tops in the forest; the whole image looks as if covered by a checkered layer. The enlargement is too large for this resolution because the image gets distorted/warped/mutilated; in other words altered. It is in this enhancement that the RAF AHB  historian recognized 'pinheads' or turrets and gun barrels whereas in fact they saw horizontal and vertical lines that were most likely added by the algorithm of a noise reduction software program which uses an overcomplete DCT dictionary (as opposed to - for example- globally trained dictionaries). This type of software denoises digital images by replacing pixels with filters from its own collection in the way that it is programmed. The image on page 38 is in all probability an enlargement of the scanned aerial photo after this type of denoising because the "grid" on it is consistent with the horizontal and vertical lines in the filters of such a dictionary.

Left: An overcomplete DCT dictionary; right: a globally trained dictionary
Source: "Hierarchical Matching Pursuit for Image Classification: Architecture and Fast Algorithms",

In reference to the tactical objects, or more specific the German tanks of the Herman Göring Fallschirmjäger Ausbildings- und Ersatz Regiment we note that on page 17, Dr. Ritchie is less sure: on the one hand he claims that they are armored fighting vehicles, on the other he concedes that they are 'impossible to identify'. Focusing on these objects (if these are moving can't of course not be discerned in a photograph) is also rather superfluous as it was only because of the 1949 annotations on the print that Dr. Ritchie's interest was drawn to this image. We might as well analyze any other object in the image.
The above validation proves that imagery analysis is not an exact science (yet) and is often a matter of interpretation. Any input on our hypothesis is appreciated by this agency.

This isn't a review of the RAF AHB publication; Dr. Sebastian Ritchie's work is much appreciated and he has undoubtedly put much time and effort in it and provides insights on aerial photo reconnaissance and image interpretation. Our investigation only provides the evidence that labeling frames 4014 and 4015 taken on 12 September 1944 as the photos that show German armor near Arnhem prior to Operation "Market Garden" is based on an incorrect assumption upon which subsequently a significant part of the RAF AHB's publication is based.
The search for the lost aerial photos of Arnhem is therefore back to the starting point.

All ten (10) aerial photos in the Royal Air Force collection at Wageningen University & Research showing the T-junction in Deelerwoud Forest, taken in 1944 and 1945.

(Click to enlarge)





Back to Case Files

(c) 2007-Present Day Email: all rights reserved.