The village of Roeux proved to be an exceedingly difficult nut to crack. On the opening day of the Battle of Arras the 4th Division had been unable to extend the gains made by the 9th (Scottish) Division, at least not to any significant degree. By the end of the 9th April our new line fell short of the road linking Roeux and Gavrelle and the gentle slopes of Greenland Hill which lay just beyond it.
Five weeks later, when Roeux was finally taken, one of the reasons why it had proved so difficult to capture became immediately apparent. The village consisted of a number of important strongpoints, almost all of which were already known to us, but what had not been fully appreciated was the degree to which these, and other locations around the village, had been linked up by an interconnecting network of underground tunnels. Once positions had been lost, these tunnels allowed the Germans defending the village to counter-attack, often enabling their detachments to emerge quickly and unexpectedly on one or more flanks, or to the rear of our troops, before the newly captured position could be consolidated. The enemy positions around Roeux consisted of very few conventional trenches, though some had been hastily improvised, often by joining up adjacent shell holes where some form of trench cover was urgently required. At the start of the battle the village had been behind the existing German trench system and had therefore not been developed as a front line defensive position, or even envisaged as such. However, the local topography, combined with the enemy’s ability to react and adapt, quickly turned Roeux into a formidable defensive position.
Before the 51st (Highland) Division’s attempt to capture Roeux on the 23rd April 1917, XVII Corps produced an intelligence summary (N0.98) covering the period from 5 pm 21st April to 5 pm 22nd April. Part of it was based on the interrogation of a junior German officer whose unit, the 2nd Machine Gun Company, 2nd battalion, 31st Infantry Regiment (18th Division), had arrived from Douai on the evening of the 9th April to take up positions around Roeux and Greenland Hill. The information provided by this prisoner would have added to the overall intelligence picture on the XVII front, particularly that part of it faced by the 51st (Highland) Division, but even more telling was what he chose to leave out of his account.
Having spent almost a fortnight in and around Roeux, the officer was clearly very familiar with its defences and the current tactical situation. He appeared to be very open about the state of morale within his unit, which he described as good, and that of the infantry garrison, which he also rated as good. He even added that the garrison was looking forward to more open fighting rather than having to endure further bombardments from our artillery. Our assessment of the prisoner concluded that he seemed reliable, but that he was not very forthcoming regarding information, noting that he considered himself very unlucky to have been captured.
Given his rank, he would probably have expected to be questioned about a variety of matters, including the state of Roeux and its defences. Rather than remain silent, which is often quite difficult to do, he offered up some information, but kept the most important details from his interrogators. Although he divulged that his unit’s machine guns were located in a forward trench, as well as in shell holes in front of it, this was largely irrelevant, not least because we already knew that the enemy’s machine-gun crews were very flexible and would readily move location to suit prevailing conditions. The trenches themselves were not really a significant factor and many of them, like those near the quarry, were only four feet deep, making them in reality little different to the many shell holes dotted about the place. The prisoner was probably correct in stating that the ones closest to the quarry were too conspicuous to be held in any strength prior to the coming battle. He indicated that four machine guns had been sent forward to a support position between Roeux Wood and Mount Pleasant Wood in anticipation of an attack, but this information, even if true, might not represent the current state of defences in that locality when it came to the actual day of the attack.
The prisoner also noted that the trenches built by his regiment’s 3rd Battalion had been poorly constructed, as a result of which the unit had sustained heavy casualties from shelling. He claimed that the 1st Battalion was also now weak in numbers, though he admitted that his own battalion had fared rather better with regard to casualties. Unhelpfully, he goes on to say that he ‘thinks’ his division ‘may’ be relieved on the 24th April by a ‘flying’ division, though he goes on to add that the relief might take place even sooner. His view is that the position around Roeux will not be given up without a fight, but that no serious counter-attack would be made to recover it if it were lost.
It should have been completely obvious to the British that the Germans had no intention of abandoning such an important position on the battlefield unless they really had no other choice. Firmly anchored to the north bank of the River Scarpe, and still linked up to the important positions around Gavrelle, it was always likely to be defended robustly. Not only that, Roeux overlooked many of our positions west of Monchy-le-Preux on the opposite bank of the Scarpe, making it even more difficult for us on that part of the battlefield. Roeux afforded very good observation over any movements between Monchy-le-Preux and the river, and if necessary it could offer supporting fire, which was something that could not be obtained directly from the German positions east of Monchy-le-Preux around Infantry Hill. Although the enemy had lost Monchy-le-Preux, the retention of Roeux would make it much more difficult for us to exploit any further progress immediately south of the river.
We also knew that fixed machine-gun emplacements were rarely resorted to unless they were located in concrete shelters, and so the information given by the prisoner was actually of very little value. Of course, like any good soldier in his position, he omitted the most important details, which crucially, on this occasion, concerned the extent of the tunnels linking the various posts and strongpoints to each other, and which also offered some measure of protection from shelling, particularly against shrapnel. He stated that he had been living in the cellar of the inn, which was located on the Roeux-Gavrelle road, and which he claimed had provided little protection from shelling, but again, this had little relevance, since piles of rubble were often just as useful to machine gunners and the infantry supporting them, particularly snipers. His information actually added very little detail to the overall picture, and hardly anything of practical value to the troops who would attack Roeux and the Chemical Works on the 23rd April.
When elements of the 51st (Highland) Division did find their way into the ruins around the Chemical Works and the area surrounding the chateau on the 23rd April, they soon found themselves under fire, often from many sides, and were quickly cut off and then eliminated or captured. A similar situation had existed on the Somme in 1916 around Mouquet Farm. The Australians had found that position very difficult to capture owing to the existence of multiple tunnels that connected one strongpoint to another. Some learning had emerged from this experience, but few preparations could be made by attacking troops unless they were aware of the existence and extent of such tunnels in the first place. Whilst it is true that the 51st (Highland) Division had also had some experience of tackling these kind of defences at Beaumont Hamel in November 1916, no two cases were likely to be the same. Without any prior detailed knowledge of the nature of the defences, and a plan to deal with them, the task of capturing Roeux was always going to prove difficult and costly. The overall casualties for the 51st (Highland) Division on the 23rd April are not disclosed in the divisional history, but we know that they were heavy. We do know, however, that the division’s total for the whole of April came to 4,496. (The Official History, Volume I, 1917 adds another hundred to the that total, but that is almost certainly a misprint)
The majority of raids that took place during the war included a brief to capture prisoners. Although many did provide valuable information, others were happy to talk, but were highly selective as to the information or the amount of detail they divulged. Nobody could blame them for this, in fact, they had a duty to give away as little as possible. Despite his capture, the German officer in question continued to serve his own side very well. However, the failure to capture Roeux on the 23rd April was only partly the result of a failure in intelligence. It failed for many other reasons that were common to the entire frontage of the attack, a front that stretched for some 10 miles, but it also failed for reasons that were particular to Roeux itself – a topic that I will cover at some later date.