by Hew Strachan
However, Stavka’s conception of the East Prussian operations was not so restricted. The balance of forces in the north encouraged hopes of a major victory rather than just a raid; furthermore, success would be defined not in terms of East Prussia alone but in terms of a Russian invasion into the heart of Germany. Although it had effectively opted for case A, Stavka’s attention in August was not only to the defeat of Austria-Hungary but also to the improvisation of a campaign against Germany. Therefore, for Grand Duke Nicholas and Yanuskevitch, the centre of gravity of the North-West Front lay to the south; its operations were to link with those of the new 9th army forming round Warsaw. In fresh instructions to Zhilinskii, on 10 August, Yanuskevitch directed the 1st army to draw the bulk of the German forces onto itself, while the 2nd army was to aim more to the west, so as to cut off the Germans falling back on the Vistula. Once the 1st and 2nd armies had cleared the area east of the Vistula, and were level with Warsaw, the 9th army would lengthen their line to the south, preparatory to the advance into Germany. To meet these instructions Zhilinskii extended Samsonov’s left to the west. Moreover, the 2nd army’s advance was no longer to be simultaneous with that of the 1st, but was to begin two days afterwards, on 19 August, to ensure that Prittwitz’s 8th German army was drawn towards Rennenkampf, so leaving its rear and communications exposed to Samsonov’s advance.
The plan was a compromise between the Front’s concentration on its initial objectives and Stavka’s concern with the second stage of operations. The consequence was the sacrifice of the Russians’ numerical advantage. At one extreme was Zhilinskii’s concern for Germany’s own proclivity for envelopment. He dissipated his strength to protect his flank and rear, particularly the areas of concentration around Kowno and Grodno. Second-line forces taken for these tasks weakened front-line strengths: up to a third of its infantry and cavalry, and two-fifths of its artillery was lost to the 2nd army.73 At the other end of the spectrum was a plan of envelopment so vast that contact between the 1st and 2nd armies could not be sustained either in space or in time. The cavalry on Rennenkampf’s left now linked up with that on Samsonov’s right. The 2nd army, with further to go, fewer roads, and a start delayed by two days, ended up five to six days behind the 1st army. Thus, the German 8th army was able to meet first Rennenkampf and then Samsonov, and to have a local superiority in manpower on both occasions. And behind it all was an abandonment of any realism inherent in plan 19: Stavka was effectively stretching its resources to implement both case A and case G, not one or the other.
The problem of conflicting priorities was not confined to one side only. Moltke’s capacity for simultaneously retaining all the options while confusing the lines of responsibility was visited as surely on the eastern front as on the western. OHL’s pre-war instructions were for the 8th army’s commander ‘to conduct operations according to his own judgement’.74 The trouble was that Moltke, with good reason, did not trust Prittwitz’s judgement; he had therefore appointed Waldersee, who was familiar with pre-war planning and with the talks with Conrad, as Prittwitz’s chief of staff. Prittwitz himself had been privy to neither. Moreover, his responsibilities were contradictory. He had to defend East Prussia; at the same time he had to do his best, with insufficient forces, to support the Austrians. Waldersee, who had been serving with the general staff in Berlin in July, knew that it might be some time before the bulk of the German forces could be freed from the west. He was also only too aware that Vienna’s focus was on Serbia rather than Russia. Germany had to encourage the Habsburg army to re-orientate itself. In the first three weeks of August Conrad’s enthusiasm for converging Austrian and German thrusts grew more pronounced.75 OHL could have interpreted this as a sign of desperation, of the Austrians’ need for the Germans to take the Russians off their backs; it could equally have been the basis for optimism, suggesting that a German attack might be sufficient to switch the Austrians off Serbia and onto Russia. On 14 August Moltke signalled to Waldersee: ‘When the Russians come, not defence only, but offensive, offensive, offensive.’76 The trouble was that by 1914 the balance of forces was even less favourable to the execution of an envelopment than it had been in 1909.
Moltke knew that. Between the Bosnian crisis and the assassination at Sarajevo more resources were allocated to the defence of West Prussia than to making East Prussia the base for an offensive. In 1912 he requested 320 million marks in order to modernize the fortifications of Posen and Graudenz, on the Vistula; he only got 200 million, but in 1913 a further 210 million (or 21.5 per cent of the total allocated to the army for that year) was also earmarked for defensive works in the east.77 The trouble was that, operationally, as general staff studies had shown in 1885, a defence on the line of the Vistula could require more men than a concentrated and directed thrust into Poland.78 Moreover, the adoption of the tactical and operational defensive was incompatible with the strategic defensive which underpinned the war’s justification in 1914: it hardly behoved the German army to begin hostilities by abandoning all Germans east of the Vistula to the Russians.
Prittwitz’s combination of apparently condoning the independence and offensiveness of his corps commanders, while choosing the Vistula as his defensive base, was thus an ineffective compromise of two fundamentally irreconcilable positions. He had not the slightest intention of being drawn too far east by Rennenkampf. He assumed that Rennenkampf’s would be the main thrust, and that the Russian 1st army—because it had the easier and more direct route—would be on him before Samsonov’s. Therefore, although he had underestimated the Russian advance from the Narew, his dispositions were sensible enough. He left a single corps, XX corps, around Ortelsburg facing south. His three remaining corps he prepared for a defensive battle on the position of the Angerapp, well back from the Russian frontier and so placed as to allow rapid redeployment to the south or retreat to the west. At the same time his dispositions did not entirely abandon the offensive, giving him the opportunity to strike Rennenkampf’s right flank from the north.
Prittwitz had correctly calculated that lengthening the Russian 1st army’s advance would diminish its fighting power. By 19 August, although Rennenkampf had only just crossed the frontier, his troops had been on the move for a week, and their rear services and supply organization were in complete disorder. Despite their considerable strength, the Russian cavalry failed to push ahead of the main body, to establish contact with the Germans or to feel round the German left flank. It had become too heedful of the virtues of defensive firepower: trained as mounted infantry, its inclinations were to dismount when opposed rather than to press forward. With Rennenkampf’s six divisions spread over a 56-kilometre front, it saw its task as closing the gaps between them. On the morning of the 19 August Rennenkampf instructed his army to rest on the 20th. The message, like many others sent by both sides in this early stage of the war, was transmitted in clear, and was intercepted by the Germans.
Prittwitz was increasingly worried by the rapidity of Samsonov’s advance. Rennenkampf’s halt would postpone the planned battle on the Angerapp, and might leave the Germans with insufficient time to cover the threat to the south. He therefore resolved to turn the headstrong actions of one of his corps commanders, von François, to good account. François commanded I corps, whose district in peacetime was East Prussia. Loath to abandon his ‘own’ territory, he had been encouraged by the general staff before the war in the idea of mounting local spoiling attacks across the frontier. By 13 August most of his corps was on the line Goldap-Stallupönen, well to the east of the Angerapp and therefore in danger of being isolated from the rest of 8th army. Prittwitz ordered XVII corps and I reserve corps to close up on the Angerapp, while I corps was to fall back on Gumbinnen: his hope was that I corps would thus be able to strike the Russians on their northern flank as they approached the main German positions on the river. François disregarded his instructions. On 17 August I corps engaged the Russians at Stallupönen. Although instructed to break off the action, François was still in contac
t with the enemy on 19 August. He reported the carelessness of the Russian advance and urged Prittwitz to allow him to attack. Prittwitz agreed. The 8th army was ordered to concentrate in line with I corps, east of Gumbinnen, and thus well forward of its planned defensive line, the Angerapp and the Masurian lakes.79
The battle of Gumbinnen reflected credit on neither side. However, it was not the flank attack planned by Prittwitz for the Angerapp position, but an operation to support I corps forced on 8th army by François’ headstrong independence. The latter’s corps began its attack at 3.30 a.m. on 20 August. It took the Russians by surprise, but given the strength of the enemy its own flanks were vulnerable. Therefore, the 2nd division switched from François’ right wing to his left, marching 16 kilometres through a forest during the night, and so pushing against the Russian right flank from 4 a.m. The Russian cavalry standing on François’s wing took no part in the action, and each Russian division was left to fight its own action against an enemy superior in numbers and in artillery. But the German attack was too hurried and too uncontrolled. On François’s right Mackensen’s XVII corps, to the south of I corps, marched forward from the Angerapp throughout the night of 19/20 August, on roads crowded with refugees, and did not enter the fight until 8 a.m. As in the battles in the west, the infantry came straight off the line of march into action without artillery preparation. The Russians had entrenched, and their artillery and infantry, linked by telephone, co-ordinated their firepower to telling effect. One Russian regiment fired 800,000 rounds from its 3,000 rifles and eight machine-guns, and its supporting field artillery discharged over 10,000 shells. The Germans, still in close order, suffered heavy losses: unable to see their enemy, they became demoralized and by 5 p.m. were in full flight; they had suffered 8,000 casualties out of a strength of 30,000.80 On Mackensen’s right, I reserve corps also had a trying night and did not make contact until 11 a.m. Its task was to strike the Russian III corps, which was facing Mackensen, in the flank. But instead it ran into the Russians’ IV corps, on the edge of the Rominten forest. Made up of recruits, and undergunned in comparison with the formations of the field army, I reserve corps closed with the enemy. The two sides were so intermingled that the Russians could not exploit their artillery and their infantry tended to fire high. But, while a tactical success, I reserve corps’ attack failed to distract the Russian III corps and so did not support Mackensen as intended.
From the German viewpoint the fighting on 20 August could not be counted a success. Three separate corps battles had resulted in incomplete victories on the flanks and defeat in the centre. However, the prospects for 21 August seemed good. Their concentration was now complete and they enjoyed a superiority of nine divisions to six-and-half. I corps had taken 6,000 prisoners in the course of the day, XVII corps had been allowed to withdraw and reorganize unhindered by Russian intervention, I reserve corps stood victorious on the battlefield, and III reserve corps would enter the battle to envelop the Russians the following day. Both François and Hoffman, the 8th army’s head of operations, were keen to finish with the Russian 1st army, the former saying that he could roll up the Russians and relieve XVII corps. Furthermore, XX corps insisted that it could hold off the 2nd army for the time being. With Rennenkampf crushed, the 8th army could with confidence face Samsonov.
But Prittwitz’s mood on the evening of 20 August was jittery. The battle of Gumbinnen was not his conception. Both XVII corps and I reserve corps had suffered heavy casualties. The fact that circumstances now presented the opportunity for a classic envelopment victory was hard to discern when the origins of the battle had been so different. Prittwitz’s inclination to break off the battle therefore developed independently of the broader situation to the south.81 But an aerial reconnaissance report that Samsonov’s army was now in Mlawa, and was therefore extending westwards and threatening 8th army’s lines of communication, confirmed its wisdom. Defence triumphed over offence in Prittwitz’s mind. His sense of caution was increased by 8th army’s exaggerated idea of Samsonov’s strength. At 7 p.m. Prittwitz ordered the action at Gumbinnen to be broken off, and the 8th army to fall back to the Vistula.
Prittwitz was doing no more than what Moltke had told him to do—to preserve his army, and to do so by falling back westwards, not by tying it up in the fortress of Königsberg. But, as Hoffman was at pains to point out, it was now too late to retreat to the Vistula. Samsonov would get there first—and would do so having effected an unimpeded concentration with Rennenkampf. If Prittwitz was to secure his line of communications, he must deal with the Russian 2nd army east of the Vistula. Hoffmann therefore proposed that François’s I corps be switched by rail to the right wing of XX corps facing Samsonov, so attacking the 2nd army’s left wing. Its entrainment would be covered by reserve units from Königsberg. I reserve corps and XVII corps should march westwards, if Rennenkampf allowed them to break contact, with a view to supporting XX corps on its left and on the Russian 2nd army’s right. But the effectiveness of this second move would depend on Rennenkampf. Osterode should be the 8th army’s area of concentration so that, screened by the Masurian lakes on one side, it could face Rennenkampf rather than Samsonov if the latter resumed his advance.
In Hoffman’s advice, given and adopted on the evening of 20 August, lay the manoeuvre which enabled the victory at Tannenberg. Hoffmann was a pre-war student of the Russian army; he had been an observer in Manchuria; he knew of the differences between Rennenkampf and Samsonov; and he was in possession of intercepted Russian messages. But, as Hoffmann himself acknowledged, the battle developed ‘entirely by itself, not as a result of a master plan.82 Indeed, this was its strength, and perhaps even Schlieffen’s real legacy. In the west the temptation to plan the campaign through to its denouement had worked against the grain of the general staff’s resources as an institution. It had made insufficient allowance for France’s reactions and, in turn, denied scope for the German army’s powers of improvisation and flexibility. In the east these very qualities were made central by the assumption of the defensive-offensive, and the consequent need for the Germans to respond to the manoeuvres of the Russians. Decisions flowed not from prescription but from intelligence and from the experience of war games and staff rides. East Prussia was the exercise area for the German army: its geography and its potential were well known to its officers. Schlieffen had recognized that the Masurian lakes created the opportunity for operations on interior lines, and that that possibility was supported by the availability of adequate road and rail links, in particular the line from Insterburg to Allenstein and Osterode. In manoeuvres in 1910 I corps and XVII corps had actually used the same communications network that they employed in August 1914. The authorship of the Tannenberg manoeuvre was therefore not confined to Hoffmann; it was institutional. At 8th army headquarters Waldersee and Grünert, the army’s quartermaster, worked to convert Prittwitz; at OHL Moltke and, at XX corps’ headquarters, Scholtz came independently to similar conclusions.83
The resolve of these officers grew on 21 August and was consolidated on the 22nd by the reports from aerial reconnaissance. Rennenkampf did not intend to pursue. This was entirely in accordance with the conclusions of the Russians’ May 1914 war game: the danger that had been revealed then was not that the advance of 1st army would be too slow but that it would be too fast. Rennenkampf’s army had expended a great deal of ammunition at Gumbinnen; it needed time to consolidate and to make good its supply position. By 23 August all the Germans facing 1st army, with the exception of a single cavalry division, had gone.84
Nonetheless, the mood at OHL was far from confident. Prittwitz’s decision to fall back to the Vistula, relayed on a bad line via Berlin, had evoked in Moltke that characteristic sense of resignation. The Vistula itself was not much of a defensive line: the water was low and it would need more men to hold it against the combined armies of the North-West Front than were available to Prittwitz. Moreover, the abandonment of East Prussia would expose Silesia to invasion and the Austro-Hungar
ian army to the full weight of the Russians. The messages coming through to OHL were admittedly contradictory. François had used his corps commander’s right of direct access to report to the Kaiser via his deputy in Königsberg. He claimed a great victory at Gumbinnen. This intelligence only confirmed Moltke’s increasing loss of faith in the command of the 8th army. Prittwitz, although now convinced of the wisdom of what Hoffmann proposed, and although the latter’s suggestions began to be put into effect on the night of the 20th, did not inform OHL or do anything to correct the impression of weakness which he had created. Equally, Moltke, despite seeing what was required—as in the west—failed to assert himself. He reckoned that only the local commander could assess the situation. He did not tell Prittwitz he was not to retreat.