By the end of September the bad weather prevalent in the North Sea had caused great dislocation in the convoy system. Ships composing convoys became much scattered and arrived so late off Lerwick as to prevent them proceeding on their passage without entering harbour. Owing to the overcrowding of Lerwick Harbour the system of changing convoy escorts without entering harbour had been introduced, and the delays due to bad weather were causing great difficulties in this respect. The question of substituting the Tyne for Lerwick as the collecting port was first discussed at this period, but the objections to the Tyne as an assembly port were so strong as to prevent the adoption of the proposal. The system of convoy outlined above continued in force from April to December, 1917, during which period some 6,000 vessels were convoyed between Norway and the Humber with a total loss of about seventy ships. There was always the danger that Germany would attack the convoys by means of surface vessels. The safeguard against such attacks was the constant presence of forces from the Grand Fleet in the North Sea. In view of the fact, however, that the distance of the convoy routes from the Horn Reef was only between 300 and 350 miles, and that on a winter night this distance could almost be covered at a speed of 20 knots during the fourteen or fifteen hours of darkness that prevailed, it will be seen that unless the convoys were actually accompanied by a force sufficient to protect them against operations by surface vessels, there was undoubted risk of successful attack. It was not possible to forecast the class of vessels by which such an attack might be carried out or the strength of the attacking force. The German decision in this respect would naturally be governed by the value of the objective and by the risk to be run. Admiral Scheer in his book states that on one occasion, in April, 1918, the German battle-cruisers, supported by the battleships and the remainder of the High Sea Fleet, attempted such an attack, but found no convoy. It was always realized by us that an attack in great force might be made on the convoy, but such risk had to be accepted. The movements of the ships of the Grand Fleet were a matter for the Commander-in-Chief, provided always that no definite orders were issued by the Admiralty or no warning of expected attack was given to the Commander-in-Chief, and, prior to the first attack on the Scandinavian convoy, no special force of cruisers or light cruisers accompanied the convoy to guard it against attack by surface vessels, although a strong deterrent to attack lay in the frequent presence of forces from the Grand Fleet to the southward of the convoy routes, which forces would seriously threaten the return of any raiding German vessels. As the enemy would naturally make the northward passage by night we could hardly expect to sight his ships on the outward trip.