by Leo Marks
SIXTY-THREE
Open Arrest
‘Stop taking their imprisonment personally. The way they’re being handled is no concern of yours … it’s Holland’s future that’s at stake.’
Nick to author, 1 March 1944
We’d been discussing SOE’s treatment of Ubbrinck and Dourlein, the Dutch agents who’d escaped from Haaren prison and reported the collapse of the Dutch Resistance to the British embassy in Berne when they’d reached Switzerland in November.
IS YOUR JOURNEY REALLY NECESSARY? was one of London’s most widely displayed posters.
Although fair play in SOE was rare and usually the result of a lapse in concentration, I was hoping that both men would be given the benefits of whatever doubts existed in the minds of those purblind enough to have any. I should have known better.
They’d arrived in London on 1 February and amplified their accounts of the German penetration, but their N-section interrogators preferred to believe the warnings from Holland that the Gestapo had ‘turned’ them and allowed them to escape to spread disinformation; they’d been sent to a holding camp in Guildford under open arrest. By early March they were still incarcerated and were likely to remain so until Giskes lost his creative flow. Nor did yet another change of leadership in N section enhance their chances of being released.
Bingham had been despatched to Australia (hopefully to the outback), and Major Dobson of the Belgian section had taken his place. Whereas Bingham believed everything, Dobson believed nothing, and his responses to Ubbrinck and Dourlein were as guarded as 84’s to the Revenue. (Three more agents had escaped from Haaren in November but Dobson was so suspicious of alleged escapees that none of them had been allowed to return to England.)
Having been given the most unenviable job in SOE (with the possible exception of mine), Dobson inevitably approached his new directorate with an eye to damage control. He was as suspicious of the traffic from Holland as he was of the escapees, and the Dutch situation was deadlocked, with the emphasis on the dead.
German aircraft were dropping bombs on London as if they were mounting a blanket attack on an indecipherable which had to be broken. It was clear that these raids were little more than target practice and that far worse was to come. As a precaution, all code records had been duplicated and stored in the country.
RF section and Duke Street had also taken action. Knowing that Tommy listened avidly to news bulletins from London and would be worried about Barbara’s welfare, they sent weekly messages to him over the BBC assuring him of her safety.
They had even more reason to be concerned about his. He’d landed at night near Clermont-Ferrand with a saboteur named Trieur, and within twenty-four hours they’d left the area and boarded a train for Paris (200 miles away). As soon as they’d reached the capital, he’d found a safe flat for Trieur and spent his first days questioning resistance-movement leaders about the latest situation in Paris, Tours and Brittany. Finding their morale low (SOE’s expanded dropping operations had only recently begun), he assured them that large quantities of arms and supplies were on their way to all parts of France, that more would be following and that the invasion would take place.
He’d discovered through his trusted friend Maud (who was on ‘friendly terms’ with a German official) that Brossolette and Bollaert were in Rennes prison and that the Gestapo still believed that Brossolette was Pierre Bourdet.
On 1 March he’d left Paris with Maud and was now in Rennes reconnoitring the prison and finalising his plans. The entire Brittany organisation had been put at his disposal for the rescue attempt.
Manuel believed it would be another two weeks before Brossolette’s dye wore off. Less of an optimist, Dismore thought it might last another one.
I accompanied Tommy on his journey to Rennes, and then embarked on a rescue operation which could no longer be postponed.
The last promise I’d made him was to simplify MOPs.
The fundamental purpose of a mental one-time pad was to make its code groups so closely resemble those of a real LOP that the enemy wouldn’t attempt to break them. And even if they learned of the subterfuge from a captured agent they would still find the code groups troublesome to break.
If the mechanics of MOPs could be simplified, they’d be stable companions to our other deception scheme (Gift-Horse), which made WOK messages look as if they’d been passed in poems to tempt enemy cryptographers to waste their precious time attacking them. The difference was that MOP messages were designed to deter them from trying.
But there was one technical problem, which was the cryptographic equivalent of getting Brossolette out of Rennes. Enemy cryptographers could identify genuine LOP code groups at a glance because of the preponderance of consonants. And those same deadly eyes could just as easily identify transposition traffic because of the preponderance of vowels. Since MOPs were based on transposition, a way had to be found to make their code groups contain the requisite number of uncommon consonants.
To achieve this, an agent would have to memorise two poems and be taught how to make a substitution square, without which no onetime pad could function. His nightmare would then begin.
He’d have to choose five words from poem A, and obtain a transposition key in the normal way. But instead of encoding a message in it, he’d encode the whole of poem B and use the resulting code group as his one-time pad! Suppose his message were:
WHOEVER DEVISED THIS FUCKING SYSTEM SHOULD BE SHOT.
If he encoded it in a poem or WOK his code groups would be:
SEOLI CTIBN SEFEI WOVEH DDTEU SEDGS UEDGS TRSEU VUKHH.
But if he used a MOP the code groups would be:
ZVKML PDYYQ XRRUV FXLLT KZPNT DSWLD APPLZ ORTTY PKHGW.
(If the latter didn’t convince the Germans that he’d used a one-time pad then I deserved to be shot.)
Putting aside the ‘deceptive element’, the system was secure enough to allow short messages to be passed in it (the shorter the better), but even twenty letters could take more than an hour to encode. I stumbled on a way to cut the time in half, but it still remained a laborious process even in ideal working conditions with no Gestapo but Nick prowling outside.
Anxious to test it on some suitable victims, I explained the system to six expert coders and three renowned plodders, and asked them to encode a lengthy MOP message.
The six experts included the phenomenal Ensign Hornung, who found nothing too difficult except being at ease. But even she had to agree with her fellow experts that the system was far too complicated and would cause endless indecipherables (including their own), and that the average coder would take for ever to learn it. They then retired to their less taxing duties.
I suddenly realised that the three plodders were still ploughing on. Although I was convinced that they had no idea what they were doing, the least I could do was wait for them to finish, and I spent the next thirty minutes pretending to be breaking an indecipherable while I watched them in silence. They finished simultaneously (a knack of plodders) and handed me their messages as if reluctant to part with them. I checked them in their presence (making three mistakes in the process) and to my astonishment discovered that every message had been flawlessly encoded.
But the miracle didn’t end there. They all volunteered to try another MOP message ‘as it was really rather fun reaching the end’.
I was tempted to hug them but consulted Heffer instead.
The Guru had no difficulty in explaining the phenomenon as he was one himself. Using the analogy of the tortoise and the hare (which he probably originated), he said that in his experience plodders were able to cope with complex processes like MOPs because their minds didn’t race ahead, and each step gave them a feeling of accomplishment, whereas experts needed to find short cuts to prove that they were experts. He added that he’d sooner be operated on by a plodding surgeon than an expert, which made me wonder which part of his anatomy had been.
Still doubting the system’s practical value, I decided to test it
on a group of briefing officers and training-school instructors, giving them no warning of what to expect.
I wasn’t surprised when they pronounced MOPs far too cumbersome to be taught to average agents in the limited time at their disposal, if at all. But one of the instructresses reluctantly agreed to try them on her present batch of pupils, who’d finished their training and were filling in time.
I wanted to sit at the back of the room to monitor her presentation but decided that my presence might put her off and that I could well be identified as the creator of the system.
She telephoned two days later to ask if I’d like her to come to London to tell me the results in person, and I at once agreed, though I suspected she fancied a shopping expedition.
Of the six agents she’d MOPped, two had given up in despair and one had gone to bed with a headache from which he had yet to recover. But the other three had decided that although it was extremely hard work, they’d like to practise the system as they wanted to use it in the field.
Astounded, I saw the agents long before I was due to in the hope of discovering what had prompted their decision.
They had no hesitation in telling me. They felt that one-time pads were so simple to use that they couldn’t possibly be safe, whereas MOPs were so complicated that no one could break them.
Recovering from my astonishment, I disabused them, and they promised to use MOPs only in emergencies. But their answer made me realise that other agents might be under the same misapprehension, and I made a mental note that in future no agents were to be considered fully briefed until they understood why one-time pads were unbreakable, with WOKs a second best.
I also decided that MOPs should be taught to carefully selected agents once their normal training was finished and that the ultimate decision as to whether to use them must be theirs.
I sensed that MOPs would have some place in the code war, if only for the wrong reasons.
On 8 March Nick left for India on a tour of inspection and Heffer took his place. Nick’s last instruction was to keep him informed about Tommy’s progress.
Messages from France reported that in mid-March Brossolette had been identified by the Germans and had been transferred from Rennes prison to Gestapo headquarters in the Avenue Foch. He’d tried to escape by jumping through a window, but had fallen five storeys and was dead. (Duke Street believed he’d committed suicide to escape further torture.)
Reports about Tommy arrived shortly afterwards. He’d returned to Paris to complete his plans for rescuing Brossolette from Rennes. Unaware that his friend was lying on a slab only a few streets away from him, he’d gone to the Paris Métro on the morning of the 21st to meet a contact and had been arrested by the Germans. He was now in the hands of the Gestapo in the Rue des Saussaies.
I prayed that he’d swallow his L-tablet before they realised who he was.
Barbara had been told of his capture.
Duke Street had informed de Gaulle.
Selborne had notified Churchill.
Merde alors, Tommy.
They cannot know
What makes you as you are
Nor can they hear
Those voices from afar
Which whisper to you
You are not alone.
SIXTY-FOUR
Misgivings
Acaptured agent’s first task, and frequently his last, was to withstand forty-eight hours of interrogation in all its forms to allow the rest of his circuit time to go underground. Tommy always warned agents that if they were caught, ‘those first forty-eight hours’ would be the hardest to endure.
He’d been in the hands of the Gestapo for seventy-two.
RF section and Duke Street suspected that he’d been betrayed by a fellow agent. They also feared that he’d been identified as Shelley.
On 10 March Delhi HQ informed Heffer that he was to remain in charge of Signals for at least a fortnight while Nick completed his inspection of India’s WT stations.
The Guru took his new responsibilities in his stroll, arriving even later than usual, leaving even earlier and exuding even more smoke rings.
His first official act was to call a meeting of all department heads to discuss our preparations for the invasion of France. The first few hours were taken up with technical wireless problems, including the allocation of new frequencies, the introduction of new signal plans and the distribution of D-Day traffic between our three stations.
I was the last to be called upon as it was believed that all was well with the code department.
Unfortunately it wasn’t.
I’d realised, hopefully in time, that there were major flaws in our contribution to Overlord and was obliged to tell my colleagues that we now had to discuss two problems which not only affected our D-Day traffic but could have repercussions far beyond it.
I was convinced that unless we took immediate precautions, the huge increase in our French traffic between now and D-Day couldn’t fail to alert the enemy that we were organising mass uprisings all over France and might even blow the imminence of Overlord.
My second worry was the use of BBC messages to agents in the field.
Taking first things first, if only for the novelty, I suggested that one way to disguise the growth of our traffic would be to start flooding the air with dummy messages, which would be transmitted round the clock to every part of Europe. Each message would be encoded in a WOK, LOP or poem and would be indistinguishable from our genuine traffic.
This minor undertaking was enthusiastically received, and Heffer allowed me forty-eight hours in which to present him with the details.
I then disclosed at length why I believed that the BBC’s en clair messages to agents had become a major security risk (responsibility for vetting them was vested in our liaison officer with the BBC).
They agreed that my misgivings were thoroughly justified and that a solution must be found as soon as possible.
Since I was the one who’d raised the problem, Heffer allowed me the same forty-eight hours in which to solve it, and I asked if I could be excused from the rest of the meeting (it seemed likely to last until well after D-Day).
Someone whispered, ‘Lucky sod,’ as I hurried from the room.
But then he didn’t know the extent of the problem I’d just talked myself into.
The idea of BBC announcers reading short en clair messages to agents in the field had been conceived in 1941 by George Bégué, the first SOE agent to be parachuted into France.
Bégué (who’d escaped from a Vichy prison in ’42 and was now Captain Noble of F section) had been given a poem-code and an elementary WT set and dropped into the Châteauroux area to communicate with London. He’d soon discovered that the Germans were jamming his traffic, that their direction-finding vans were scouring the vicinity and that he was risking his life every time he came on the air. He’d also realised that many of London’s messages consisted of instructions to carry out orders he’d already been given. Anxious not to use his set if he could possibly avoid it, he’d suggested to London that their last-minute instructions to him could safely be conveyed in short prearranged phrases, whose meaning only he and F section would know. If London agreed, he would listen every night to the BBC’s foreign service until these phrases were broadcast.
His concept of ‘personal messages’ was at once adopted and rapidly spread to every country section in SOE. Since then, short plainlanguage messages had become an integral part of agents’ communications and were currently being used to confirm safe houses, passwords and dropping operations, substantially reducing an operator’s air time.
They also fulfilled a function which Noble hadn’t foreseen. They enabled agents in the field to say to those whose help they badly needed but who doubted their bona fides, ‘Make up a short message – it doesn’t matter what – and I’ll arrange for it to be broadcast a week from now on the BBC’s foreign service.’ The results of this offer never failed to produce the desired effects and often enabled agents to borr
ow large sums of money on the lender’s assumption (not always well founded) that London would repay the advance when the war was over.
These all-important phrases had been christened ‘iodoforms’ by someone in Baker Street with a classical education, though it was hard to guess his identity.
A typical iodoform was ‘Je ne regrette rien’, and a typical Signals problem was that the agent it was intended for had to be aware of its significance; the only way London could convey it to him was through WT messages which he’d then have to acknowledge.
There was no danger in this if WOKs and LOPs had been used, but if the details were transmitted to him in a poem-code (as so many iodo-forms were) the agent was likely to have beaucoup to regret. If the enemy had broken his poem, they’d know the meaning of his ‘personal messages’ and be in a position to take appropriate action.
And there was another danger: even if they hadn’t broken his code, if they scoured his traffic for the words of iodoforms they suspected were his, their anagrammers would have a field day, and the life of the poem would soon be over. And the agent’s with it.
But an even greater nightmare was rapidly taking shape.
I’d been told by Nick that shortly before D-Day SOE intended to use iodoforms to instruct secret armies and agents that it was time to break cover and cause maximum havoc in every way they could. I’d been too preoccupied with recruitment problems to point out that a) even on D-Day dozens of agents would still be using poem-codes, either because we’d failed to deliver silks to them or because they’d mislaid them, and that b) if the D-Day iodoforms had been prearranged in broken poems, the enemy would not only know that Overlord was imminent, they might be able to pinpoint where the landings would take place.
I didn’t need forty-eight hours to decide what had to be done.
But I wondered what else we experts had missed.