Sunday, 7 June 2020

The Arnhem Air Reconnaissance Puzzle

Over many years, the Arnhem air reconnaissance story has been dominated by a single account of how an RAF Spitfire photographed German tanks near the city shortly before Operation Market Garden.

Reference: Arnhem: The Air Reconnaissance Story

This blog examines the subject of air reconnaissance in Operation Market Garden, the Allied airborne assault on The Netherlands, mounted in September 1944 in an attempt to capture Arnhem and cross the River Rhine - the last natural barrier protecting Hitler’s Germany. The operation ultimately failed, not least because the Allies underestimated the scale and speed of the German response. A key part of that response came from 2 SS Panzer Corps - 9 and 10 SS Panzer Division - which had been sent to the Arnhem area to rest and refit.

The role of 2 SS Panzer Corps in the Arnhem story has inevitably led historians to ask the question - did Allied intelligence know they were there? Eventually, it became clear that some intelligence had reached Allied commanders suggesting that German armour was in the area. This included signals intelligence - so-called ULTRA - and reports from the Dutch resistance. But the most famous part of the Market Garden intelligence story is concerned with air reconnaissance, and claims that German armour was actually photographed near Arnhem shortly before the operation was launched.

This claim was made by the former British Airborne Corps head of intelligence, Major Brian Urquhart. Urquhart recalled that a single low-level mission succeeded in capturing imagery of German tanks, assumed to belong to II SS Panzer Corps, and his account attracted particularly close attention after the publication of Cornelius Ryan’s book, A Bridge Too Far, in 1974. Both the mission and its aftermath were then subject to highly emotive dramatisation in the film of the same name.

Urquhart recalled harbouring deep misgivings about the Market Garden plan, which seriously underestimated the likely strength of German opposition in The Netherlands, in his view. But he failed to convince the British Airborne Corps commander, Lieutenant General FAM ‘Boy’ Browning. He therefore requested a low-level air reconnaissance mission, which was flown by a Spitfire based at RAF Benson. Urquhart’s aim was to obtain photographic evidence that German tanks were positioned at Arnhem. The Spitfire allegedly executed ‘a low-level sweep’ across the Arnhem area, and Urquhart was soon afterwards supplied with five photographs showing tanks and armoured vehicles parked under trees. In later correspondence with the historian Martin Middlebrook, Urquhart identified the tanks as Panzer IIIs and Panzer IVs. A former brigade major with 1st Airborne Division, Anthony Hibbert, also recalled that Urquhart had shown him photographs of Panzer IVs parked beneath trees.

Yet Browning was still unmoved. There were no subsequent changes to the Market Garden plan, and Urquhart was soon afterwards ordered on sick leave on the grounds that he was suffering from nervous exhaustion.

The Spitfire mission as depicted in A Bridge Too Far

A low-level image of tanks on the edge of woodland, prepared for
A Bridge Too Far; this is what everyone expected to see, but
the reality was very differen
In the film, A Bridge Too Far, Urquhart was renamed Major Fuller, but the film otherwise sought to recreate the events he described as faithfully as possible. The RAF Spitfire is shown flying at ultra-low level, and the photographs provide an oblique-angle view that clearly captures tanks and armoured vehicles at the edge of woodland. Major Fuller’s subsequent attempt to make Browning see reason must rank as one of the most memorable scenes of the entire film and has exerted a profound influence on the historiography of intelligence in Market Garden ever since.

Unfortunately, however, the series of events Urquhart recalled has proved extremely difficult to document. Several extensive searches for the photographs failed to locate them. Urquhart insisted that the Arnhem sortie was flown by a Spitfire squadron based at Benson; this would almost certainly mean 541 Squadron, part of 106 Group, which had a particular geographical responsibility for The Netherlands. Yet the UK National Collection of Air Photographs contained no oblique imagery showing tanks at Arnhem.

In addition, although the Benson missions were systematically recorded at squadron and group level, not one record matched the sortie Urquhart described. Low-level missions targeting the Arnhem, Nijmegen and Grave bridges on 6 September were scrupulously noted down, their photographs meticulously archived, but all other recorded reconnaissance sorties over Arnhem were flown at higher altitude and captured vertical imagery. Equally, it proved impossible to locate an interpretation report derived from a low-level mission that photographed German armour near Arnhem before Market Garden.

It must be acknowledged here that relatively few  tactical interpretation reports have survived in the UK archives. Nevertheless, historians were left to draw their own conclusions, and these might be summarised as follows. Either the photographs and all documentary evidence of their existence had been destroyed, whether deliberately or through the routine weeding process, or Urquhart’s recollections were not entirely accurate. A related mystery concerned the whereabouts of the tanks on 17 September 1944, the date Market Garden was launched. Had German tanks been positioned near the city that day, they would have been deployed immediately in response to the airborne assault, yet no Panzer IIIs or IVs were encountered in Arnhem until the 19th, and it is known that these tanks were transferred from Germany following the Allied landings. What, then, became of the tanks in the photographs?

This long-running puzzle has now finally been solved chiefly by focusing on the process by which Allied air reconnaissance missions were requested, planned, and executed in 1944. Consideration of this process quickly raised doubts about the precision of Urquhart’s account.

The wartime documents record that, to capture oblique imagery at low level, 541 Squadron Spitfires were equipped with wing-mounted forward-facing synchronised 8-inch lens cameras. As these only produced photographs of a limited area, the target location had to be established and briefed to the pilot before the mission was flown, and the aircraft had to be flown directly towards the target when the photographs were taken.

The camera configuration used by 541 Squadron Spitfires
to capture oblique imagery at low level in 1944
Consequently, this technique was reserved for fixed points of interest. Examples include radar stations, harbour facilities and munitions depots, and the sorties that photographed the Arnhem and Nijmegen bridges on 6 September. By contrast, of the few low-level missions executed by 541 Squadron in the summer of 1944, not one was launched to photograph tactical targets, such as mechanised ground formations. Obliques of mobile ground targets were captured by tactical reconnaissance squadrons based on the continent, but with rear-mounted sideways-facing cameras fitted in aircraft that flew parallel to target areas - not towards a specific pinpoint. The 541 Squadron Spitfires did not use this camera configuration.

It is profoundly unlikely that 541 Squadron would have obtained any photographs of tanks in the Arnhem area on 12 September 1944 without at least some prior guidance - a point I shall return to later. However, while a high-level mission might have located tactical targets using intelligence about a general area, the low-level approach would have required advance information identifying the exact location of the enemy formation at some kind of fixed facility, such as a barracks. Yet there is no recorded evidence or claim that German tanks had previously been spotted at a specific geographical position.

Low-level oblique images of the Arnhem and Nijmegen bridges taken 
by 541 Squadron on 6 September 1944; the squadron reserved this 
technique almost entirely for fixed points of interest
This raises the question of whether the mission, as described, would even have been officially sanctioned, for it would have involved considerable risks without much chance of operational gain. The documents note that all requests for air reconnaissance tasks involving the 106 Group squadrons had to be approved by a body named the Joint Photographic Reconnaissance Committee (JPRC). This was a powerful committee that met twice per day to assess requests for photographic cover and prioritise between reconnaissance tasks. The JPRC had also to clarify poorly-worded or ill-judged applications and had a particular responsibility for clarifying the geographical coordinates of air reconnaissance objectives.

To put it bluntly, the JPRC would never have acceded to a request for a low-level air reconnaissance mission to search some general area around Arnhem for German armour, when the proposed task would have been carried out by a squadron that had no established low-level capability against mobile tactical targets and was only equipped to take low obliques of fixed points of interest. The probability of mission success in such circumstances would have been minimal.

So what did 541 Squadron do when they were required to photograph smaller, tactical targets? There is no great mystery about this because they spent a great deal of time searching for V2 launchers in the later months of 1944. In fact, when seeking such targets without prior knowledge of their location, the squadron’s approach was overwhelmingly to operate at high altitude using cameras with 36-inch lenses to take vertical imagery of areas, rather than pinpoints. The task of finding tactical objects in the photographs was left to the RAF’s photographic interpreters (or PIs). Consequently, it is very likely that the mission recalled by Brian Urquhart was flown at higher rather than lower level and employed a conventional, vertical, camera angle.

A 541 Squadron Spitfire; cameras used to take conventional vertical
images are visible on the rear under-side of the fuselage

The first RAF photograph of a V2 rocket being prepared for
launch - a high-level vertical image
The significance of this point is hard to exaggerate, but another vital lead emerged when the photographic interpretation process was considered in detail. It transpired that there was a system for communicating very high priority intelligence to customers, known as the Form White. If an initial examination of imagery at RAF Benson revealed information of particularly high importance, a Form White might be raised and dispatched before the film was sent to the Allied Central Interpretation Unit at Medmenham (ACIU) for more detailed analysis. If the airborne forces had requested a mission over the Arnhem area, which obtained photographs of German tanks, it seemed highly likely that a Form White would have been produced and sent to Urquhart.

Although the forms themselves have not survived, their creation is shown in the records of the ACIU. In the period before Market Garden, Spitfire sorties in the prospective battle zone led to the production of only a handful of Form Whites. One of these was mission 106G/2816, flown by Flight Lieutenant Brian Fuge on 12 September 1944, and this became a particular focus of attention. Apart from the fact that it was flown by the right squadron (541 Squadron) on approximately the right date, and was entirely confined to the airborne objectives, it also  resulted in the generation of a single Form White relating to the Deelen area, a few miles north of Arnhem.

The ACIU sortie records, showing 106G/2816
and the Form White referring to Deelen

Flight Lieutenant Brian Fuge
Unfortunately, when imagery from this mission was requested from the National Collection of Air Photographs (NCAP), it transpired that none was held in the UK archives or any other known location. A further request for imagery from other sorties covering Deelen on 12 September was ultimately rewarded by a wartime plot of photographs taken by a 544 Squadron Mosquito. A single frame on to the plot (4023) over a forest known as the Deelerwoud had been marked with the letter ‘A’, and this letter would also have appeared on the photograph itself (or more probably an enlargement of part of the photograph), and on an accompanying interpretation report. However, frame 4023 was also missing from the NCAP collection, and there was no sign of the interpretation report among the (very incomplete) files at the National Archives (link to air reconnaissance conservation and curatorial policy).

A wartime map plotting photographs taken by a 544 Squadron
Mosquito over Deelen on 12 September 1944; note the 'A'
marked on frame 4023
At this stage, there seemed to be no obvious way forward. Further progress was impossible until the Dutch air reconnaissance archives were placed online. It then transpired that the imagery from mission 106G/2816 had survived. It was included within a quantity of RAF film gifted to the Dutch government at the end of the Second World War. An online search for photographs therefore began, focusing on exactly the same location as the Mosquito frame 4023.

Two photographs taken shortly after the Spitfire camera was activated - frames 4014 and 4015 - were found to have covered this area, and frame 4015 had been marked with several letters. At first, these were assumed to refer to objects of military significance described in an interpretation report. In fact, the Dutch images, like those in Britain, were ‘master copies’ that would not have been marked in any way during the war. All visible annotations on the photos held in The Netherlands were added later. Nevertheless, by an extraordinary coincidence, high-resolution copies of both frames revealed a large column of military vehicles, including armoured vehicles, either parked near an intersection or moving along a track through the Deelerwoud, and several larger rectangular objects, which were suspected of being supply dumps (link to full photograph and mapping).

Mission 106G/2816, frame 4015, showing the German armoured
column and objects believed to be supply dumps
Assuming a connection between the column and the supply dumps, one obvious question arose. Why were the vehicles not parked immediately next to the dumps? The answer is quite straightforward. The Deelerwoud is largely planted with tall conifers bearing only a few high branches. They would not have provided very good cover. However, near the intersection, the track is lined with deciduous trees with far more dense foliage. They are clearly visible in the wartime imagery and - much enlarged - still stand by the track to this day. The German tanks were parked in this area partly because there was room for them by the track, but also because it provided the best available concealment from Allied air power.

The Deelerwoud's conifers are widely spaced and
would have provided only partial cover

The deciduous trees at the track intersection,
now much larger than in 1944
In seeking to identify the vehicles in the photograph, the key challenge could be summed up in one word - resolution. The technique used to enlarge air photographs in the Second World War involved producing blow-ups directly from the negative of the frame rather than the print. This provided the maximum clarity and detail. To give some idea of the clarity that resulted from this approach, a wartime enlargement of three Panzer IIIs is reproduced here.

Note also, certain key features of the Panzer III illustrated by this photo. One is the forward-positioned turret. But observe the appearance of the upright frontal armour and the front of the hull, too. From the air, we see two parallel lines and a circle.

Another illustration of this enlargement process is provided by imagery of German anti-aircraft batteries in the Arnhem area shortly before Market Garden. These enlargements would have represented a small fraction of area images captured by aircraft flying at high altitude, yet they depicted the two batteries in extraordinary detail.

In the absence of negatives for the wartime photographs, it is impossible to recover the resolution that would have been available to Allied PIs in 1944. In short, we will never be able to view frame 4014 or 4015 in the same detail that would have been available to them because of the loss of resolution that occurs during the transition from negative to print, and then from print to electronic scan.

The two frames were taken in extremely sunny conditions, and there was a great deal of light reflection from the top of the vehicles (although there was also a lot of shadow and tree cover). Nevertheless, the resolution was so poor that it was initially hard to determine the orientation of the column. At first, due partly to the position of the most visible tank turret, it seemed as if the column might be moving north. The common German practice of rotating turrets during servicing and maintenance was not taken into consideration until accumulating evidence demonstrated unequivocally that the vehicles were heading in the opposite direction.

Panzer IV maintenance, turrets turned; note also the raised rear
hatch, the shadow of which can be seen behind the turret shadow
in 106G/2816 frames 4014 and 4015
Some useful help came from the Air Ministry’s wartime guidance for photo-interpreters. In addition, the War Office’s Air Recognition Manual drew attention to certain specific features of the Panzer III and IV that appeared in vertical imagery. For the Panzer III, these included the ‘turret set slightly forward’. For the Panzer IV, the attention of interpreters was drawn to the ‘cupola at extreme rear and centre of the turret, which is set midway’ and ‘wings protruding beyond main body’. Further relevant information from the Air Ministry guidance concerned the appearance of German supply dumps, which, it was said, 'normally consist of stacks, often about 20ft square, sited in woods or along tree-lined roads'.

The most obvious approach available to study the appearance of the tanks involved digital enlargement and enhancement of the imagery to the maximum extent, accepting some pixelation. The objects in the photograph could then be compared with further relevant imagery from the wartime period, drawing on the Air Ministry and War Office guidance. Via this means, it was possible to define the basic rectangular outline of tank hulls, such items as tank turrets, upright frontal armour, and the Panzer IV ‘wings protruding beyond main body’.

In addition, while light reflection obscured some smaller objects, it revealed others in extraordinary detail. The identification of a Panzer III stemmed not only from the visibility of its turret and gun and the forward position of the turret (relative to the Panzer IV), but also from the symmetrical pattern of rear hatch hinges on top of the hull.

A Panzer IV parked under the trees left the markings of a tank track on the ground, perfectly aligned with the wing of the stationary vehicle. Beyond this, the length-width ratios of the three most visible tanks accorded with the Air Ministry interpretation guidance.

No less revealing were the common features of the various objects in the photograph. Given the limited resolution available, a single speculative tank turret identification might reasonably have been called into question. However, at least four turrets are visible, and vertical frontal armour can be seen in at least three cases.

Relative to one another, these features were also positioned in accordance with the general layout of early model Panzer IIIs and IVs. The rear hatch hinges of the Panzer III (labelled 'A' below) were located behind the turret. The front of the hull of the suspected Panzer IV under the trees (labelled 'C') was positioned in front of the vertical frontal armour. This tank, and the Panzer IV (labelled 'D'), clearly displayed the correct relative positioning of the frontal armour and the protruding front wings. The turret of the Panzer IV labelled D was correctly positioned in the middle of the hull, well behind the frontal armour, just as the cupola was correctly positioned at the back of the turret. The tank (probably a Panzer III) on the other side of the track (E) displays first the hull front, then the upright armour, then the turret. Additionally, the measurable distance from the vertical frontal armour to the rear of the hull was identical for tanks C and D, supporting the identification of tank C as a Panzer IV.

So, while there was insufficient resolution and too much cover from trees or shadow to provide more than a few reliable identifications, it was clear that the larger tanks included Panzer IVs of early design, with short-barrelled 75mm guns; there were also smaller tanks, including Panzer IIIs. These were likewise early models equipped with 37mm guns.

A lone AFV heading south into the Deelerwoud, pictured 
in 106G/2816 frames 4014 and 4015
Another AFV was visible near the intersection - possibly a half-track with a short-barrelled rear-mounted gun. This might have been an SDKFZ 250/8 or 251/9. Further up the track, moving through the woods, was yet another turreted AFV, which was impossible to identify. Of the other vehicles visible in the photograph, none was readily identifiable. Still further to the north, a second AFV with a rear-mounted turret or gun was visible in the open, possibly stationary, and more vehicles could just be seen emerging from the trees. A diagrammatic representation of what Allied PIs might have seen at the track intersection in frames 4014 and 4015 is shown at the following link with tank identifications shown as certain (C), probable (PR) and possible (PO) 

Further imagery of this area was captured on 19 September. This revealed extensive scarring on the ground left by parked vehicles, but no military equipment was in evidence. Equally, while the 12 September imagery clearly showed several vehicles moving down the track through the Deelerwoud, no such movement was in evidence on the 19th.

Comparative imagery showing (left) 12 September and (right) 19 September 1944. The armoured column is clearly visible on 12 September; the 19 September image only shows
scarring on the ground left by parked vehicles.

Who, then, did the tanks in the Deelerwoud belong to? Were they under the command of 9th or 10th SS Panzer Division? The reconnaissance battalion of 9th SS Panzer Division, under Hauptsturmfuhrer Viktor Graebner, was positioned at the village of Hoenderloo, a short distance north of Deelen, before Market Garden. Indeed, the German vehicles visible in the photograph were parked or moving along a track that ran directly south from the village. The battalion did possess a number of SDKFZ 250s and 251s, but there is no record that they were equipped with early type Panzer IIIs or IVs or that such tanks were involved in the Arnhem battle. More modern Mk IIIs and IVs only arrived from Germany on 19 September, two days after the airborne landings began. Probably, then, the armoured unit in the photographs was not part of II SS Panzer Corps.

There was, however, another formation in this area of The Netherlands that possessed a considerable number of older tanks. This was the Hermann Goering Parachute Panzer Training and Replacement Regiment, which was responsible for supplying replacements to its parent division - the Hermann Goering Parachute Panzer Division, then fighting in the East.

The regiment’s Second Battalion was tasked with training panzer, panzer-grenadier, self-propelled artillery and self-propelled anti-tank gun personnel. Based at Utrecht, it is recorded that they had previously used Apeldoorn and other locations north of Arnhem for training purposes, and a 1 Parachute Brigade intelligence summary prepared on 13 September 1944 confirms that the Allies were well aware of this fact: ‘Before last June, the area Arnhem - Zwolle - Amersfoort was an important training area, particularly for armoured and motorised troops, including SS and Hermann Goering reinforcement units.’

In the first week of September, the Hermann Goering Parachute Panzer Training and Replacement Regiment became part of the 1st Parachute Army, formed under Generaloberst Student with the aim of constructing a defensive line on the Albert Canal to block the British advance from Antwerp. Soon afterwards, elements drawn from the Training and Replacement Regiment were sent south. They suffered heavy losses fighting at Hechtel between the 7th and the 10th but managed to extricate at least some tanks.

On the 11th, the day before the Spitfire mission, all remaining units were ordered to move from their base areas to Eindhoven. As their commanding officer noted in his diary, ‘Even the recruits are to be sent in. Otherwise, there is nothing more available.’ What remained of the Second Battalion was positioned north of Eindhoven. It is known that they were equipped with early model Panzer IIIs and IVs, as these tanks were encountered near Son by 101st Airborne Division soon after the first Market Garden landings on the 17th.

A Panzer III of the Hermann Goering Parachute Panzer Training
and Replacement Regiment, knocked out by Allied aircraft north
of Son on 17 September 1944

A Panzer IV of the Hermann Goering Parachute Panzer Training
and Replacement Regiment in The Netherlands

A possible scenario is that at least part of the Second Battalion was held in the rear as a reserve during the fighting at Hechtel - perhaps the less battle-worthy of their six companies - and was training north of Arnhem when orders for the move came through. It is plausible that, in preparation, they refuelled and restocked from the dumps near Deelen, where they were photographed by Flight Lieutenant Fuge's Spitfire.

Again, it should not be thought that this south-western movement of German rear units escaped detection by Allied intelligence. On 14 September, First Allied Airborne Army’s head of intelligence noted:

'The main factor, on which all sources agree, is that every able-bodied man in uniform who can be armed is in the battle – the Germans are desperately short of men and it is improbable that any formations capable of fighting will be found in an L[ine] of C[ommunication] area, however important it may be ... Identifications in the Albert Canal area satisfactorily prove that practically all the enemy troops which could have been in Northern Holland are now actually engaged.'

The movement of any available Hermann Goering elements from the Arnhem area to Eindhoven is entirely consistent with this assessment.

*          *          *

If the Spitfire mission examined in this blog was unrelated to the sortie recalled by Brian Urquhart and immortalised in A Bridge Too Far, it would mean that two very similar events occurred at almost exactly the same time, one of which has featured in almost every published work on Operation Market Garden (although it lacks any primary documentation), while the other has been entirely hidden from history until recently.

This is wholly implausible. While British Airborne Corps intelligence undoubtedly requested Flight Lieutenant Fuge’s mission and presumably received the Form White after he landed, Brian Urquhart only ever recalled one occasion when German armour was photographed near Arnhem. It might just be that, on the basis of the imagery, he managed to arrange a low-level sortie that was not subject to the normal request and approval processes and which was never recorded in the squadron or group records. However, this seems equally unlikely, given the lack of low-level tactical reconnaissance expertise at RAF Benson, the intense pressure imposed on the squadrons by the V-2 search and potential objections on the grounds that the enemy vehicles would have moved elsewhere by the time the sortie was flown.

Far more probably, the five photographs supplied to Urquhart were enlargements showing parts of two high-level vertical images, frames 4014 and 4015, which clearly depicted the ‘tanks and armoured vehicles parked under trees’ that he and Anthony Hibbert recalled, and the Panzer IIIs and IVs that they specifically identified. The supposition that the imagery was captured at low level may perhaps have stemmed from the scale of the enlargements, for the German vehicles would have been difficult to see unless the photos were blown up to the greatest possible extent. Alternatively, Urquhart may have requested a low-level sortie that was not, ultimately, flown; or he may, looking back, have confused this episode with the earlier 541 Squadron tasking to photograph the Arnhem, Nijmegen and Grave bridges.

Contrary to popular belief, the photographs did not confirm the presence of 2 SS Panzer Corps near Arnhem shortly before Market Garden. Rather, they showed tanks of obsolescent early wartime design belonging to the Hermann Goering Parachute Panzer Training and Replacement Regiment, tanks that are known to have relocated near to Son by 17 September 1944 and which were encountered there on that very day by 101st Airborne Division. 

Two associated issues merit further historical analysis. The first is operational security. If the RAF had flown a high-altitude mission over Arnhem in September 1944, this might have suggested to the Germans an interest in an array of military and economic entities extending from barracks and anti-aircraft batteries to rail transport. However, a low-level sortie along the Neder Rein at Arnhem, as flown on 6 September, could only have targeted the bridges, and the specific Allied interest in bridges in the area would have been confirmed by the other two sorties flown that day over Nijmegen and Graves.

This was potentially a considerable risk. When concerns over the operational security of Market Garden were subsequently raised within First Allied Airborne Army, they made specific reference to ‘bridges that we photograph but do not bomb’. If, as seems likely, the wisdom of mounting the 6 September missions at low level was later called into question, this may well explain why all other air reconnaissance activity over Arnhem was conducted at high altitude.

The second issue concerns the speed with which mission 106G/2816 captured imagery of tanks near Arnhem. According to Brian Urquhart’s account, the 541 Squadron Spitfire was dispatched for the explicit purpose of photographing German armour. Flight Lieutenant Fuge’s fourteenth frame fulfilled this remit. To all intents and purposes, he flew straight out to the Deelerwoud, banked his aircraft, activated his camera and, within a few seconds, photographed a column of German armoured vehicles.

The intelligence process is like a jigsaw puzzle. It is rare for particular 'nuggets' to be acquired independently or without some form of guidance. More typically, intelligence involves the painstaking assembly and fusion of interrelated inputs gathered over time in a variety of ways. Therefore, while it is not suggested here that the Allies already knew exactly where the German armour would be, it is hard to avoid the suspicion that this famous Spitfire sortie benefited from other information, and, if so, it would be interesting to know its source. There may yet be another twist in the Arnhem air reconnaissance puzzle that would add significantly to our knowledge of intelligence gathering during the prelude to Operation Market Garden.