A reconstruction of the Andrea Doria/Stockholm collision

By Carl O. Nordling


In his book Collisions and their Courses, Richard A. Cahill has clarified the fundamental causes of the Andrea Doria/Stockholm collision in 1956.

There can be no doubt that the officers in command on both ships deliberately took some improper risks. Consequently, they have rightly been blamed for the disaster that followed.

The taking of risks does not, however, in itself cause a collision; it only enhances its probability. We may still ask what was the immediate cause of the impact. The aim of this article is to answer that question. The procedure will be a detailed analysis of the movements of the vessels the minutes before and during the collision.

It is quite clear from Mr. Cahill’s investigation that the testimonies of the responsible persons are contradictory. Therefore, these cannot be relied on in the context of the critical happenings.

Instead, the following analysis will be based primarily on "hard" facts. That is to say, on physical remains that were examined after the event by several persons. Among these remains are the damages of the hulls, the SOS-signal received from Andrea Doria after the collision, and the preserved course recorder charts from both ships. Oral evidence (produced in court) will be used only to establish the speed of the ships, the adjustment of their clocks and their CR (Course Recorders). Also the passing distance between Andrea Doria and the lightship Nantucket Shoals (and its point of time) will be obtained from testimony.

According to the uncontradicted statements made by the responsible officers, it will be assumed that the speed of Andrea Doria was 21.8 knots (11.2 m/s) and that of Stockholm 18.5 knots (9.5 m/s).

It will also be assumed that the CR of Andrea Doria was set 9 hours ahead of GMT (Greenwich Mean Time) and its angular recording 11° too high, while the CR of Stockholm recorded its time 2 minutes ahead of GMT and points of the compass 2° too high. These assumptions are in agreement with other evidence and are necessary in order to explain how the ships could be where they were when they met.

The effects of the collision on the hulls of the ships

The photographed damages on the hull of Stockholm include total destruction of the forebody until 22 meters from the stem. Andrea Doria is reported to have received an indentation on its starboard side reaching 9 meters inwards from the plating. The destroyed material that became squeezed together must have formed a buffer, with an estimated thickness of about 3 meters. With this assumption, the total penetration of Stockholm turns out to have been about 28 meters (i.e. 22+9-3 meters).

After having penetrated thus far, the forward velocity of Stockholm had been retarded to zero (from an initial 8 to 9 m/s). We may therefore estimate that the penetration took about 7 seconds. After the penetration, Stockholm retracted itself about 6 meters (28-22) from the hull of Andrea Doria. This movement may have taken another 7 seconds or thereabout.

During the phase of penetration, part of the energy of motion of Stockholm was transmitted to Andrea Doria, forcing the latter towards port and increasing its port sheer slightly (the pressure being applied forward of the center of gravity). At the same time, part of the energy of motion of Andrea Doria was transmitted to Stockholm. As a result Stockholm must have turned rapidly to starboard and dragged through the water as well. Thus Stockholm acted as a break (or drag-anchor) on Andrea Doria, reducing the speed of the latter markedly. We will therefore assume that the speed of Andrea Doria was only 15.5 knots as an average during the 10 minutes that ensued after the contact with the hull of Stockholm. Consequently Andrea Doria would have covered about 2.6 nautical miles during these 10 minutes.

The position of the place of collision

At 0320 GMT on July 26 the following message was sent from Andrea Doria by wireless: "SOS de ICEH here at 0320 GMT pos. N 40.30, W 69.53, need immediate assistance." (Position determined with LORAN according to officers on duty.) There is reason to regard the reported position as fairly accurate since accuracy was essential for the guiding of rescuing ships to the then unmanœuverable Andrea Doria. The present position of the wreck is N 40°29’30", W 69°51’00", only 1.6 nautical miles from the SOS position. Before the sending of the SOS-signal, Andrea Doria had made two sheers, the first 110° to port during tree minutes, the second about 90° to starboard during the following seven minutes (according to the CR). Assuming that Andrea Doria’s speed during these ten minutes was 15.5 knots and that she steered a course of 264° at the beginning of the first sheer, her position at that moment would have been 1.1 n. miles to the north and 0.7 n. miles to the east of the SOS position. That would mean that the 110° sheer to port began at position N 40°31’10", W 69°52’00", with an estimated margin of error of about 0.8 n. miles. As we shall see soon, the collision occurred very close to this point.

The possible effect of a current

Andrea Doria apparently drifted less than two n. miles during the 11 hours that elapsed from the SOS till she sank to the bottom. This indicates that there cannot have been much of a current in the waters where the collision took place. Where would Andrea Doria have been at 0310 GMT according to dead reckoning from her last known previous position? She passed about one n. mile south of the lightvessel Nantucket Shoals at 0220 GMT. In 1956 the lightship was situated at N 40°33’00", W 69°28’00". As the liner passed, her course was altered from 261° to 268°. This course was kept during 45 minutes, until 0305 GMT, when it was changed to 264°. The latter course was steered in five minutes until the 110° sheer began. The said courses and an assumed speed of 21.8 knots would have brought Andrea Doria to the position N 40°31’10", W 69°52’00" at 0310 GMT. This is exactly the same position as the one arrived at by counting backwards from the signaled SOS position. Consequently there is no reason to reckon with any current at all in connection with the collision.

Some distance to the west of the area of the collision there was indeed a current causing Stockholm to drift northwards. The commanding officer made several adjustments during the evening in order to compensate for this drift. At 0230 GMT he determined Stockholm’s position by means of bearings to three radio beacons (on Block Island, Nantucket Island and Nantucket Shoals respectively). The position was N 40°31’10", W 70°08’30" with a margin of error of about 0.4 n. miles. If there had been no current, Stockholm would have reached the position of N 40°30’40", W 69°52’00" at 0310 GMT. This is half a mile farther south than the above calculated position of Andrea Doria. In reality, the two ships must have been very close to each other at 0310 GMT. Therefore, there might have been a current deflecting Stockholm northward even after 0230 GMT. However, the margin of error inherent in all these calculations suffices to account for the difference of half a mile between N 40°31’10" and N 40°30’40".

The direction of Stockholm at the moment of impact

It has been generally assumed that a certain blot on the CR graph of Stockholm was caused by the first physical contact between the two ships. There is, however, no solid reason for this assumption. The first impact of the bows affected the rest of the ship very little and was barely perceptible at the bridge. The crushed sheeting served as a shock absorber protecting the main part of the ship from violent retardation. No blot was found on the CR graph of Andrea Doria although its CR was much closer to the point of contact. The blot is probably either accidental or the result of a slight doctoring of the graph before it was made public. And, before all, there is positive proof that the collision did not begin when Stockholm was steering the course indicated at the point of the blot (i.e. 130°).

At 0310 GMT the CR of Stockholm recorded a sheer of 60° to starboard within 20 seconds. A 60° sheer would have taken at least 180 seconds if caused by the rudder and engine only. Therefore this sheer must necessarily have been caused by some external influence. The only possible influence seems to be the onhooking of Stockholm to Andrea Doria and the resulting rotatory movement of the former. According to the CR graph, Stockholm was steering a course of 150° at the beginning of the 60° sheer (and 210° after completing it). That is to say that Stockholm was steering 150°, and not 130°, at the moment of impact. The blot at 130° does not indicate the impact.

The direction of Andrea Doria at the moment of impact

The destruction of the prow of Stockholm reached a little farther from the bows on the port side than on the starboard side. The photographs show three missing letters of the ship’s name on the port side against only two on the starboard side. The distance between letters was one meter. From this fact we may deduce that the angle between the two ships was about 85° at the moment when the penetration ceased. Supposedly, this happened about seven seconds after the first impact. The rotative movement during these seven seconds was about 20° according to the CR graph. Therefore the angle between the two ships must have been about 105° (or between 100° and 110° considering the margin of error). Since Stockholm was steering 150° at that moment, it follows that Andrea Doria was steering a course of about 255° ( 5°) and that she had consequently managed to turn about 9° to port after having followed the course of 264° during five minutes. (The officers of Andrea Doria estimated that she had completed 10° to 15° of her port sheer at the moment of impact.) This turn of about 9° took some 15 seconds, during which time the ship moved 0.09 n. miles forward. Consequently Andrea Doria’s position at the beginning of the collision, "point Zero", would have been N 40°31’00", W 69°52’04" according to dead reckoning.

The position of the ships when they observed each other

The position of both ships may now be calculated for all the critical moments during the half-hour preceding the collision. This can be accomplished by means of dead reckoning backward from "Point Zero" according to the known speeds and the recorded courses.

About 25 minutes before the collision, Stockholm was at a point 7.53 n. miles to the west and 0.67 n. miles to the north of "Point Zero". At the same time Andrea Doria was at a point 9.07 n. miles to the east and 0.45 n. miles to the north of "Point Zero". The true bearing of Stockholm as observed form Andrea Doria would have been 2.7° on the starboard bow. Because of the inexactitude of the radar outfits of the 1950’es, the echo may have appeared at any bearing between 1.7° and 3.7° on the starboard bow. (The officers of Andrea Doria stated that they had seen the echo at bearing 4° on the starboard bow at a distance of 17 n. miles, i.e. 47 minutes before reaching "Point Zero".)

About 15 minutes before the impact, Stockholm was at a point 4.6 n. miles to the west and 0.61 n. miles to the north of "Point Zero". The simultaneous position of Andrea Doria was 5.4 n. miles to the east and 0.32 n. miles to the north of "Point Zero". The echo of Andrea Doria should have appeared on Stockholm’s radar screen at a bearing between 0.2° on the port bow and 1.8° on the starboard bow. The officer in charge testified that he had seen the echo at a bearing 2° on the port bow, i.e. about 3° from the calculated true bearing.

About nine minutes before the collision the echo should have appeared at a bearing between 1.3° and 3.3° on Stockholm’s starboard bow. The officer in charge maintained that he had seen it at 4° on the port bow at this moment. His testimony can hardly be given credit, since it is too much at variance (more than six degrees) with the substantiated facts.

About five minutes before the collision, the echo of Stockholm should have appeared at a bearing of between 7° and 9° on the starboard bow of Andrea Doria. The officers of the latter stated afterwards that at this moment the echo had been observed at bearing 15° on the starboard bow. Nor can this statement be given credit (being 7° at variance with the calculated true bearing).

Stockholm’s disastrous blunder

If a regular and secure plot had been made from any two observations between 25 and 5 minutes before the collision, it would have shown a CPA (Closest Point of Approach) of about 0.4 n. miles (750 m) starboard to starboard. Any competent officer in charge on an Atlantic liner should have regarded such a CPA condition as absolutely intolerable. However, no plot at all was made on Andrea Doria, and her 4° sheer to port at this ort distance was clearly insufficient. It would have increased the (irregular-sided) CPA to a mere 0.55 n. miles. This must still be regarded as too narrow, especially when approaching in a fog or starboard to starboard. As it was, both these conditions were at hand.

The officer in charge on Stockholm stated that he had indeed plotted the echo of Andrea Dora using a "Bial Plotter". He said that his plot was based on the two echo points at bearings 2° and 4° on the port bow. The true bearings, however, must have been about 0.8° and 2.3° on the starboard bow. The inexactitude of the method used may at a pinch explain that real angles of 0.8° and 2.3° might have appeared as 2° and 4° respectively--but they would certainly still have appeared on the starboard bow. The behavior of the officer confirms that he really must have believed that the approaching vessel was on Stockholm’s port bow. It is extremely improbable, however, that the radar outfit should have inverted its picture starboard to port and vice versa. This kind of confusion is much more common in the human brain. As a matter of fact, most people have occasionally mistaken right for left or vice versa. Then there are a few individuals suffering from an outspoken ambilateralism, i.e. a real difficulty to tell right from left from one second to the next one. Any such ambilateral person would of course be disqualified as a ship’s officer from the beginning. But certainly there are all kinds of intermediary forms between the disability and the normal occasional disturbance. Some individuals may fall victims to serious ambilateralism only occasionally, perhaps just a few times in a lifetime. Such an individual may not know about his handicap, and he would not be barred from any trade.

Such an isolated lapse into ambilateralism seems to be the most probable explanation of the otherwise unaccountable behavior of the officer in charge on the bridge of Stockholm. He stood on his starboard to starboard course until his ship was only 2.5 n. miles from the expected CPA. Then he ordered a starboard sheer, although the approaching ship was on the starboard bow, just as the radar echo must have been all the time.


This is the kind of accident that is bound to happen at times even if it is extremely rare. No rules can stop the human brain from mixing up right and left once in while. This analysis of the detailed course of events in the Andrea Doria/Stockholm collision therefore emphasizes the importance of Mr. Cahill’s words: "Where sea room is available, a prudent mariner does not let another approach so close, that some sudden and unanticipated move does not allow time to take effective evasive action."

We may add that when incommunicable vessel approaches on an opposite course, one should never make way by moving to port. Either the course should be kept (if the expected value of CPA permits) or else a significant move to starboard should be made in good time. And if the worst comes to the worst, one should try to use the forebody of the ship as shock absorber--not the vulnerable broadside.


An attempt at explaining the behavior of Stockholm’s Third Mate (Johan-Ernst Carstens-Johannsen) before the collision with Andrea Doria and his recollection of what he had seen.

1. What we see is to a great extent formed by what we expect to see. That is to say, we tend to interpret our impressions in a way harmonizing with the general situation that we expect to have around us.

The Third Mate probably was used to steering at a point several miles south of the Nantucket Shoals lightship and to adjust the course nearer to it only when in sight. This seems to have been the custom. He would also have expected ships arriving from the Atlantic to pass either North of the lightship or rather close to it on the south side. Consequently, it would have been natural for him to expect to meet ships port to port in the area now in question. (Actually, Stockholm happened to be steering a course that would pass very close to the lightship.)

2. When the Third Mate made his first observation of the radar echo of Andrea Doria it was 10 nautical miles afar. As he remembered it, he would have seen the echo 1° on the port side of longitudinal axis of the ship. At this very moment the direction of the axis is said to have been 90°. The reconstruction shows that the real bearing of Andrea Doria must have been about 0.8° on the starboard bow, assuming that Stockholm’s real course was 90° (as the course recorder chart indicates). In real life, the course was not steady but oscillated between 88° and 92° (90°-94° as the course recorder had it). The Third Mate’s first impression may therefore have been an echo placed 1° degree on the port side of the axis (if his ship happened to head 92° at that very moment). This first impression of an echo on the port bow may have taken root in his mind. Therefore, when he later saw the echo in other positions, he may just have contented himself with measuring its distance from the axis--as he already "knew" on which bow it was. Since he had ordered the course 89° and since he considered the echo to be 1° to the port of 90°, he was now compelled to adjust the bearing 1°, which would have meant zero. He should have considered the echo to be exactly in the course line of Stockholm. But since he "knew" that the echo was on the port side he may unconsciously have made the correction in the wrong direction and added one degree to the radar bearing instead of subtracting it. Thus he could have put the echo to be 2° to the port of Stockholm’s course.

3. When the radar echo showed the approaching ship to be nearly two miles distant, the Third Mate may have realized (perhaps only vaguely) that something was wrong. This insight (or feeling) may have been the reason why he ordered his first starboard sheer, changing Stockholm’s course from 90° to about 110°. Already before this sheer was completed and when the ships were still about 1.6 nautical miles apart, Andrea Doria would have appeared on Stockholm’s port bow--if it chanced to come out of the fog at this very moment. The first faint impression of the lights must have been difficult to interpret. In such a situation it is natural to "see" what one expects to see, i.e. a ship on the port bow showing a red light. A few moments later, when the sight became clear and sharp, it was certainly obvious to all that Andrea Doria showed its green light.

4. It is, as rule, extremely difficult to remember the correct order of happenings occurring in rapid succession during a stressful situation. When the Third Mate and his look-out both say that Stockholm’s starboard sheer started after Andrea Doria was seen, their memory may have been influenced by so called rationalizing. It must have seemed rather natural to think that the appearance of Andrea Doria caused the sheer. Otherwise, the sheer would have appeared rather gratuitous.

5. Generally, we tend to interpret our own behavior as logical and just if there is any possibility at all to do so.



Cahill, Richard A., Collisions and their Causes. London 1983.

Coast Guard’s 1956 List (for position of Nantucket Shoals lightship).

Mattson, Algot, Den Långa Natten. Göteborg 1986 (for position of Stockholm)

Moscow, Alvin, Collision Course. New York 1959 (for CR graphs, etc.)

Nautical Chart No. 2860 (Outer Approaches to New York, scale 1:500,000)

Notice to Mariners No. 32 of August 11, 1956, Part I (for position of wreck).


Carl O. Nordling