In the Battle of the Atlantic, any U-boat that transmitted on short-wave gave away a bearing, and an Allied station that heard it could draw a line on the chart toward the boat. High-frequency direction finding, called HF/DF or Huff-Duff, worked from shore stations around the ocean and later from the mastheads of convoy escorts. This piece covers four things: how the equipment took a bearing on a transmission lasting only seconds, how several bearings became a fix, how the Germans tried to shorten their signals, and why the popular story that gives Enigma decryption all the credit leaves out a separate and major contributor.

Hunting the null: why loop direction finders missed short transmissions

Radio direction finding was already a mature technique by 1939. Ships and coast stations had been using rotating loop antennas for navigation since the First World War. A loop receives most strongly when its plane points toward the transmitter and least when it is broadside to it. To take a bearing, the operator turned the loop by hand while listening, found the point where the signal faded to a minimum, and read the bearing off a scale. Operators worked on the minimum, called the null, because it is much sharper than the maximum. On a steady medium-frequency beacon the method gave good results, but it needed the signal to stay on the air while the operator swung the loop back and forth across the null to confirm it. A transmitter that keyed for a few seconds and stopped could leave the operator with a rough direction and no real bearing.

U-boats sent their long-range traffic to U-boat command ashore on high frequencies, because short-wave signals bend back from the ionosphere and return to earth at great distances. That skywave path caused a second problem for a loop. A signal coming down from the ionosphere arrives at a steep angle with a horizontally polarized component, and its angle and polarization shift as the ionosphere moves. The horizontal parts of a loop pick up that component, so the null blurs, wanders, or disappears. Operators called the effect night effect, because at medium frequencies it grew worse after dark, and it made loop bearings on distant HF transmitters unreliable even when the operator had plenty of time.

Histories of the campaign generally describe the German navy as knowing that Allied shore stations took bearings on U-boat transmissions. They also describe it as assuming that a short HF message would end before anyone could take a usable bearing, and that bearings from distant stations would carry errors too large to be of tactical use. Both assumptions matched the limits of the loop, and the Allied equipment that went into service had been built to get around those limits. Another article in this section covers what U-boat callsigns, schedules and traffic patterns gave away to Allied analysts. This one stays with the equipment, the geometry of a fix, the German countermeasures, and the question of who deserves the credit.

Adcock arrays and a cathode-ray tube: how the instantaneous set worked

The wartime HF/DF set combined two older ideas. The first was an antenna arrangement patented by Frank Adcock shortly after the First World War. An Adcock array replaces the loop with spaced vertical elements connected in pairs. Vertical elements respond to the vertically polarized part of an incoming wave and largely ignore the horizontal part, which removes most of the polarization error that ruined loop bearings on skywave signals. The array still has the directional pattern needed for direction finding, because the difference between the signals at the two elements of a pair depends on the angle at which the wave arrives. Two pairs set at right angles give two outputs, and together those outputs carry the bearing.

The second idea came from Robert Watson-Watt and his colleagues, who worked between the wars on locating thunderstorms from the radio crackle that lightning produces. A lightning stroke lasts a fraction of a second, so a rotating loop was useless for that work. Their answer was to feed the outputs of two crossed antenna systems through two matched receiver channels to the two sets of deflection plates in a cathode-ray tube. When a signal arrived, the spot on the tube was pushed in two directions at once and traced a line whose angle matched the bearing of the source. The operator read the bearing from a scale around the tube face as soon as the signal appeared, and the display needed no hunting and no swinging of the antenna.

The engineering problem was keeping the two receiver channels matched in gain and phase while they were tuned across a wide band of frequencies, because any mismatch tilted the trace and introduced error. A trace drawn straight through the center of the tube also has a 180-degree ambiguity, since the same line points to the transmitter and directly away from it. Sets added a sense antenna, a simple omnidirectional element whose signal could be combined with the array output to show which end of the line was correct. The British shipborne sets usually named in histories are the Admiralty FH3 and its successor, the FH4, and American escorts carried their own HF/DF sets. I describe the operating principle they had in common and leave the model-by-model detail to specialist histories of the equipment.

Ground wave at the masthead: what one bearing told an escort

A U-boat transmitting from close by reached an escort’s receiver by ground wave, the part of the signal that travels along the surface of the sea, and not only by skywave. The ground wave from an HF transmitter dies out within a few tens of miles over salt water, the exact distance depending on frequency and power. Inside that range it gives a strong, steady, vertically polarized signal, and an Adcock set reads a sharp, stable bearing from it. A skywave signal from a distant boat sounds and looks different, because it fades and its trace on the tube wanders. A trained operator could judge from the character of the signal whether a transmitter was inside ground-wave range, and inside ground-wave range meant the boat was close to the convoy.

That judgment answered the question an escort commander cared about most. One bearing from a ship’s own set, together with the knowledge that the transmitter was near, gave him a direction in which to send a ship at once. Short range also made a small bearing error matter very little. At 15 nautical miles, a 2-degree error puts the boat about half a mile to one side of the true line. A destroyer or corvette running down that line could reach the area while the U-boat was still on the surface, where boats of the period normally had to be to transmit, and either sight it or force it to dive.

A ship is a poor platform for direction finding. The masts, rigging, funnels and other metal near the antenna re-radiate the incoming signal and bend the apparent bearing, and the error changes with the bearing. Escorts mounted the HF/DF array as high as they could, usually at the masthead, so that it stood clear of most of the ship. Each installation was calibrated by swinging the ship through a full circle against a transmitter at a known bearing and recording the correction at each angle. A set that had been fitted but not calibrated could give bearings that looked confident and were wrong by enough to send an escort past the boat.

Crossing the lines: how bearings from several stations became a fix

One bearing gives a line of position, meaning the transmitter is somewhere along that line. A bearing from a second station in a different place gives a second line, and the transmitter lies near the point where the two cross. A third bearing should pass through the same point. In practice, because every bearing carries some error, three lines usually form a small triangle, which navigators call a cocked hat, and its size shows the operator roughly how much to trust the fix. A large triangle meant at least one bearing was bad, or the transmitter was far enough away that small angular errors had grown into large distances.

Distance multiplies angular error. A 1-degree error at 1,000 nautical miles puts the line about 17 miles to one side of the true position, and skywave bearings on a boat in mid-ocean could easily be off by several degrees. The angle at which the lines cross matters as much as the size of the error. Lines crossing near a right angle give a compact area of uncertainty. Lines crossing at a shallow angle stretch that area into a long thin shape along the lines, so a fix from two stations close together, both looking out into the ocean, could be a strip many tens of miles long. Shore networks were spread around the Atlantic for this reason, with stations in Britain and on the far side of the ocean in Canada and the United States, so that bearings from opposite sides would cross at useful angles.

Shore bearings went to the operational intelligence plots that tracked U-boats. In Britain that was the Submarine Tracking Room in the Admiralty’s Operational Intelligence Centre, run for much of the war by Rodger Winn. The US Navy had a corresponding Atlantic Section in Washington under Kenneth Knowles, and Canada ran its own plot at naval headquarters in Ottawa. A shore fix in mid-ocean was too coarse to send an escort to a boat, but it was good enough to show where boats were concentrating, and the plots used that picture to route convoys around waiting patrol lines. Shipborne sets did the close work. Two escorts in the same screen, each with HF/DF, could take simultaneous bearings on one transmission and produce a cross-bearing fix within a few miles of the convoy.

The birdcage on the mainmast: putting HF/DF on convoy escorts

Most histories place the first British shipborne HF/DF sets at sea around 1941, with fitting spreading through 1942 and into 1943 until many escort groups had at least one equipped ship and often two. The antenna was usually the open frame at the top of the mainmast that sailors called the birdcage, and it competed for the highest position on the ship with the new centimetric radar, which also wanted a clear view. Where a ship’s layout allowed, designers moved masts and fittings around to give both a reasonable outlook. A specialist operator stood watch on the set, searched the bands U-boats were known to use, and reported every bearing and his assessment of ground wave or skywave to the bridge.

The tactical value came from the way the wolf packs worked. Under the pack system, the first boat to find a convoy was expected to report it to U-boat command and to keep in contact while other boats were directed toward it. Those contact and shadowing reports were exactly the short transmissions that shipborne HF/DF had been built to catch. When an escort picked up a ground-wave signal, a ship could run down the bearing and force the shadower under. A submerged boat of that period was slow, and with its periscope as its only view it soon lost touch with a convoy that was making turns. The rest of the pack then lost the reports it needed to close in, and many convoy actions ended with no U-boat sunk and few merchant ships lost.

That kind of result is easy to undercount, because a pack that never managed to attack leaves very little in the record compared with a sinking. HF/DF bearings could also warn an escort commander that several boats were transmitting ahead of the convoy before any attack began, which gave him time to ask for air cover, change course, or reposition his screen. Escort commanders who learned to trust their operators used the set as an early-warning sensor as well as a weapon for running a boat down. HF/DF reached the escorts alongside other new tools, including centimetric radar, the Hedgehog ahead-throwing weapon, escort carriers and very long range aircraft. Accounts of individual actions usually credit several of these together.

Short signals and Kurier: the German effort to beat the bearing

The German navy’s main answer to direction finding was to shorten its transmissions. U-boats used short-signal books that reduced common reports, such as a convoy sighting with its position, course and speed, to a few coded groups, and a separate short code for weather reports. A message that once took a minute or more of keying could go out in a small fraction of that time. The thinking matched the loop-era assumptions described earlier, because a shorter signal does give a hand-swung loop less time to find its null. Against an Adcock set with a cathode-ray display, which produced a bearing as soon as the signal appeared, shortening the message reduced the risk much less than the Germans expected.

The short-signal books connected the two Allied intelligence efforts in an unexpected way. On 30 October 1942 men from HMS Petard boarded the sinking U-559 in the eastern Mediterranean and recovered short-signal material, including the weather short-signal book, before the boat went down. Two of the boarding party, Lieutenant Anthony Fasson and Able Seaman Colin Grazier, went down with the boat. Histories of Bletchley Park credit that material with helping the codebreakers resume reading the four-rotor Atlantic U-boat traffic in December 1942. A document created partly to defeat direction finding ended up helping to break the cipher.

The German navy also restricted transmissions and, in later phases of the campaign, ordered more radio silence near convoys, which cost the pack system much of the coordination it depended on. The technical answer was Kurier, a burst system that recorded a message in advance and sent it as a single compressed burst lasting a fraction of a second. Published descriptions of the burst duration vary, and I do not give a figure. Histories generally describe Kurier as developed and tested in 1944 and fitted to only a few boats in the last months of the war, too late to affect the campaign. Accounts indicate that Allied intelligence knew about the system, but I have not seen a reliable account of how Allied direction finders would have performed against it in regular service. Histories also generally conclude that the Germans attributed many detections to radar and underestimated shipborne HF/DF for most of the war.

Enigma and HF/DF: what the record says each contributed

The breaking of the German naval Enigma ciphers at Bletchley Park was one of the great intelligence achievements of the war, and nothing in this section takes away from it. The problem is with the popular version of the story, which became dominant after the Ultra secret became public in the 1970s and often presents decryption as the reason the Atlantic campaign turned. The record supports a different picture. Decryption and direction finding were separate sources that answered different questions, the tracking rooms combined them with every other source available, and each covered gaps the other left.

The clearest evidence comes from 1942. On 1 February 1942 the Atlantic U-boats switched to a four-rotor Enigma machine on a new key, and Bletchley Park could not read that traffic for most of the year, until December 1942. Convoys still sailed during those ten months, the tracking rooms still plotted shore HF/DF fixes, and escorts still ran down ground-wave bearings, with direction finding and traffic analysis supplying most of the picture of where the boats were. Both sides were reading the other’s traffic at the same time. Most histories state that the German naval intelligence service, the B-Dienst, read the Allied convoy cipher known as Naval Cypher No. 3 until it was replaced in June 1943, which gave U-boat command convoy routing information of its own.

When Ultra was available, it often told the Allies what U-boat command intended, such as patrol line positions, refuelling rendezvous and orders to particular boats. Decrypts could arrive hours or days after the signal was sent, depending on how quickly a given day’s key was broken. HF/DF reported only that a transmitter was on the air at a given bearing, but it did so at the moment the boat keyed, and on an escort that was often what decided the encounter. Kathleen Broome Williams’s book-length study, Secret Weapon: U.S. High-Frequency Direction Finding in the Battle of the Atlantic, published by the Naval Institute Press, is one detailed treatment of the case for direction finding’s importance. For the record, see the official naval histories and the published accounts of the tracking rooms.

Locating a transmitter today: what the Atlantic says to radio shops

Public safety radio shops still have to find transmitters. A stuck microphone on a conventional repeater channel, interference from a faulty device, a 121.5 MHz homing signal from an emergency beacon, or an unauthorized user on a channel all present the same problem the escorts faced, which is working out where a signal is coming from with only what the signal reveals. On a trunked or digital system the console often shows the unit ID of the transmitting radio, and that does part of the job without a bearing, because it identifies the radio even though it does not locate it. On an analog conventional channel with no ID signalling, a stuck transmitter is anonymous, and someone has to go out with direction-finding equipment and take bearings.

The geometry has not changed since 1942. One bearing gives a line, two give a fix, and a third gives the size of the error. Bearings taken from points close together produce a long, thin area of uncertainty, while bearings from widely separated points crossing near right angles give a compact one. Reflections near the receiver distort bearings much as a ship’s superstructure did, so a bearing taken beside a metal building or a vehicle deserves less trust than one taken in the open. The Civil Air Patrol and amateur radio direction-finding enthusiasts work with these problems regularly, and many counties have one or both nearby without the radio shop knowing it.

The Atlantic also has a lesson for tactical users. The U-boat commanders assumed that a brief transmission was safe because they judged it against equipment they knew about. A modern tactical team carries radios, phones, and sometimes GPS-reporting devices that may transmit automatically, and the team leader may not know what each device emits or who could be listening. That is a planning question for the radio administrator and the team, and it stays at the awareness level here. The credit question carries over as well. An after-action review that gives a good outcome to one tool, when four sources contributed, will lead the agency to fund and train the wrong things, in the same way the popular Enigma story still leads readers to overlook direction finding.

What to do at your agency

• The radio system administrator should key a test radio on each conventional channel and each talkgroup this month, record whether the dispatch console displays the transmitting unit ID, and write the list of channels with no ID into the existing stuck-microphone procedure as the channels that would need physical direction finding.
• The communications center manager should identify, by name and phone number, the nearest Civil Air Patrol squadron and any local amateur radio group that does direction finding, and add both contacts to the existing interference and emergency beacon section of the dispatch SOP.
• The radio shop supervisor should add one paragraph to the existing interference-hunting procedure stating that bearings are taken from at least three points spread well apart, away from vehicles and metal buildings, so that the lines cross at wide angles.
• The tactical team commander should ask the radio administrator for a written list of every device the team carries that transmits automatically, including GPS-reporting radios and phones, and review that list at the next scheduled team training day.
• The QA or training officer should revise the after-action review template so that it asks reviewers to list every source of information that contributed to the outcome, and use the revised form at the next scheduled review.
• Whoever maintains the agency’s training or history slides should search them for any claim that Enigma decryption alone won the Battle of the Atlantic and correct it to say that decryption and direction finding were separate sources the Allies used together.

Takeaways

• Hand-turned loop direction finders needed a signal that stayed on the air while the operator found the null, and skywave signals blurred loop bearings, so short HF transmissions were hard to locate with older equipment.
• Wartime HF/DF combined Adcock antenna arrays, which reject much of the polarization error in skywave signals, with a twin-channel cathode-ray display that showed a bearing the moment a signal appeared.
• A shipborne set could tell from a ground-wave signal that a U-boat was within a few tens of miles, which gave an escort both a bearing and a reason to run it down immediately.
• Bearings from two or more stations crossed to form a fix, and the accuracy of the fix depended on distance, bearing error and crossing angle, which is why shore stations were spread around the Atlantic.
• Forcing a shadowing U-boat to dive often broke up a pack attack even when no boat was sunk, and that kind of outcome is easy to undercount.
• The German navy shortened its signals with short-signal books and later developed the Kurier burst system, which histories describe as reaching only a few boats too late to affect the campaign.
• Enigma decryption and HF/DF were separate and complementary sources, and during the 1942 blackout on the four-rotor U-boat key, direction finding and traffic analysis supplied most of the picture.
• Published percentages that assign U-boat kills to a single source depend on how the counting was done and should not be repeated without the source and its method.