Clearance and Provenance

Foyle had been at the file since seven, which was when he preferred to begin any serious piece of work, and by half past nine he had established, to his own satisfaction, that the operational chronology in the cover summary was wrong by at least three days regarding the initial contact protocol with DOVETAIL-7.

The error was not trivial. Three days, in the context of a running operation in the German Democratic Republic in the spring of 1974, was the difference between a controlled approach and a scramble. Whoever had compiled the summary — and the initials at the foot of page two belonged to a junior in Registry whose name Foyle could not now recall, though the handwriting suggested a woman, small and forward-leaning, with Continental sevens — had dated the first meeting between Harbird and the asset from the ninth of March, when in fact the cable traffic from Bonn Station placed it firmly on the twelfth. Foyle made a note in his ruled pad, small and precise, and turned to the next document.

His study occupied what the estate agents had once called the second reception room, a designation that had amused Vivian and irritated Foyle in roughly equal measure when they purchased the house in 1982. The room was south-facing, which in summer made it too warm for sustained concentration and in February made it merely tolerable, and he had arranged it as he had arranged every workspace since his first desk at the Annexe in 1966: reference materials to the left, active file at centre, notepad to the right, lamp positioned to eliminate shadow across the reading surface. The banker’s boxes on the shelving unit against the far wall were labelled by year in his own hand, spanning 1968 to 1997. Three of the labels had been rewritten; the originals had faded beyond legibility. He kept meaning to go through and redo the lot but the task required a free afternoon and a certain administrative will that he found harder to summon since his retirement, though he would not have put it in those terms.

The DOVETAIL file had come to him through a route he preferred not to examine too closely. Officially, operational records from the period remained under restriction, but portions of the file had been declassified under the 2003 review, and what had not been declassified had been, as it were, insufficiently secured during the physical migration from Century House to Vauxhall Cross, when certain filing cabinets had sat in a loading bay for the better part of a week under the supervision of contract removal men who possessed neither clearance nor curiosity. Foyle was not a thief. He had simply, through the ordinary courtesy of former colleagues who understood the value of institutional memory, acquired copies of documents to which his former clearance grade would have entitled him access, and which in any case concerned an operation in which he had been a principal desk officer. The distinction between possession and access mattered to him, though he recognized it might not matter to everyone, and he had taken the precaution of storing the copies in a locked filing cabinet rather than mixing them with the declassified material. Procedure. Even now. Especially now.

He was looking, specifically, for the decision chain that had led to Thomas Harbird’s premature recall from Bonn Station in April of 1974. This had been, in Foyle’s long-held and carefully considered view, the catastrophic error of the DOVETAIL operation — the removal of the case officer at the precise moment when the asset, one Rudolf Kirschwasser (cryptonym DOVETAIL-7, clearance STRAP 2, handling restricted to indoctrinated personnel), was signalling escalating concern about his own operational security. Harbird had been handling Kirschwasser for eleven months. The rapport was established. The communication protocols were tested and functioning. The dead drops were clean. To pull the case officer at that juncture was, Foyle had argued at the time and continued to believe with the quiet certainty of a man who has had three decades to examine his reasoning, an act of managerial cowardice by Gerald Innes, then Head of the Europe Controllerate.

Innes had been dead for six years — a stroke, at his house in Devon, where he had retired to keep bees and vote Conservative with the uncomplicated satisfaction of a man whose conscience had never troubled him about anything more serious than the wine list. His reputation, among those few who still remembered him at all, had settled into the comfortable amber of institutional nostalgia — good man, safe pair of hands, not imaginative perhaps but sound, which in the vocabulary of the Service meant he had never done anything bold enough to be blamed for. Foyle did not share this assessment. He had watched Innes manage DOVETAIL from a position of chronic caution, hedging every operational decision, routing every risk upward through committees and sub-committees until the operation existed in a permanent state of bureaucratic suspension, alive on paper but comatose in practice. When the crisis came — when Kirschwasser began reporting possible surveillance and a potential penetration of Bonn Station’s contact procedures — Innes had panicked. He had pulled Harbird. He had left the asset hanging in the wind without a handler for three weeks. And by September, Kirschwasser was in a Stasi interrogation facility in Hohenschönhausen, and Innes was writing damage-limitation reports that read like press releases for his own competence. He received a commendation the following year. Foyle had seen the citation. It praised his “prudent management of an operational contingency.” Prudent. The word still produced, in Foyle, a sensation that a less controlled man might have called contempt.

This was the narrative Foyle had carried for thirty years, refining it in retirement the way another man might refine a handicap or a fly-casting technique — patiently, with a craftsman’s attention to detail and a craftsman’s pleasure in the work. It was clean. It was consistent. It was supported by the evidence as he remembered it. And it placed the failure precisely where Foyle believed it belonged: on the desk of a man who had lacked the nerve to run the operation he had authorized. What the file would provide, Foyle was confident, was the documentary confirmation — the routing slips, the cable traffic, the circulation lists that told you who had seen what and when and in what order — that would allow him to set the record straight. Not publicly, of course. He was not that sort of man. But a paper, carefully prepared and scrupulously documented, lodged with the Histories Section for the benefit of future researchers — a correction to the institutional memory. A matter of accuracy. Nothing more.

On page forty-seven of the file, between a carbon copy of a liaison report from Bonn Station and a traffic analysis summary that someone had annotated in red pencil, Foyle found the routing slip for Harbird’s recall.

It was a standard Personnel Movement Authorization, Form C-17, the pale blue paper gone nearly white at the edges and soft with age. The request was for the immediate transfer of T. J. Harbird from Bonn Station to London Station, effective 12 April 1974, citing “operational reassessment (priority pending)” as justification. The circulation list showed three desks. The first was the Personnel Section duty officer, who had stamped and countersigned. The second was the Europe Controllerate registry clerk, whose initials Foyle recognized — Pamela something, a mousy woman who had later transferred to GCHQ. The third desk — the originating authority, the one that had drafted and initiated the form — was his own.

The signature at the bottom was clear. A. R. Foyle. In his careful, slightly backward-leaning hand.

Foyle looked at the signature. He turned to the next document in the file. Then he turned back. He examined the form number, the date stamp, the distribution list, the carbon register. The form had been filed in the correct sequence. It bore the correct authorization stamps. His signature was exactly as he would have signed it in 1974 — the distinctive loop on the capital F, the compressed final e that Vivian had once said, early in their marriage, made his name look as though it were trying to leave the page before it was finished being written.

He made a note: C-17 routing anomaly — my signature on originating desk. No record of instruction from Innes. Possible verbal authorization not minuted. Check whether diary for April ‘74 survives.

There was no diary for April 1974. He had established this months ago, at the beginning of the review. The pocket diaries from 1973 to 1976 had been among the personal effects he had surrendered on retirement — standard procedure, nothing sinister, a condition of the pension and the ongoing obligation of confidence. No matter. The signature on the form was certainly his, but a signature on a routing form was not a decision. It was an administrative act. Someone had instructed him to initiate the transfer, and that someone, within the chain of command as it existed in April 1974, could only have been Innes. The fact that no minute existed recording the instruction was regrettable but not remotely unusual — fully half the operational decisions in the Europe Controllerate had been made in the corridor between the third-floor offices, over a whisky, or in the margins of a meeting about something else entirely, and never committed to paper. That was how the Service worked. The paper trail was the skeleton of the decision, not its flesh.

He turned the page.

The next significant document was a letter, handwritten on unlined paper in a slanting, educated German that Foyle recognized immediately as Kirschwasser’s. His own German had been adequate for operational correspondence but not for the finer textures of nuance and feeling, and he was grateful for the translation stapled behind the original. The letter was from Kirschwasser, addressed via dead drop to his case officer, dated 3 April 1974 — nine days before the recall order that bore Foyle’s signature. It ran to five paragraphs, and its tone, even filtered through the translator’s flat institutional English, was unmistakably that of a frightened man working to control his fear:

I have reason to believe that the current meeting arrangements are known to hostile elements. On two separate occasions in the past week I have observed what I believe to be directed surveillance at the secondary contact location (the bookshop on Argelanderstrasse). The surveillance was not consistent with routine Stasi neighbourhood monitoring — the individuals were stationary, positioned with sight lines to the entrance, and departed when I did not enter. I do not believe this is coincidental. I request an immediate review of all contact protocols and consideration of emergency extraction procedures as discussed in our contingency planning of November 1973.

Foyle remembered this letter. He remembered it arriving on his desk via the overnight diplomatic bag from Bonn, on a Tuesday morning — he could not recall the exact date, but Tuesdays were when the bag arrived, and he had been in the habit of processing its contents first thing, before the morning meeting. He remembered reading it with the careful attention he gave to all incoming asset correspondence. And he remembered, with perfect clarity, his assessment at the time: that Kirschwasser was prone to alarm. That the asset had flagged similar concerns twice before — once in October 1973, regarding a suspected approach by a BfV officer at a scientific conference, and once in January 1974, concerning anomalous foot traffic near his apartment that had proved, on investigation by Bonn Station, to be a Bundespost survey of post-box locations. In both cases the reported surveillance had been either imagined or benign. Kirschwasser was a competent source — his technical intelligence on Warsaw Pact signals infrastructure had been rated among the best product from the Eastern Bloc in that period — but he was a nervous man, and nervous men saw watchers behind every lamppost.

Foyle had initialled the letter, filed the original in the operational chronological, and sent a note to Bonn Station recommending that the secondary contact site be reviewed at the case officer’s discretion.

He had not forwarded the letter to Innes. He had not flagged it as priority traffic. He had not requested an operational stand-down or the activation of Kirschwasser’s extraction protocol. This had been, in his professional judgment, the correct response. An alarmist report from a source with a documented history of false positives did not warrant escalation to the Head of Controllerate. It would have been unprofessional. It would have been, in the precise language of the operational guidelines that Foyle had helped to draft in 1971, a disproportionate response.

He found the note in the file — his note to Bonn Station, typed on the standard buff memorandum paper — and read it again now with the mild satisfaction of a man recognizing his own best work. The note was well-drafted. Measured. It acknowledged the concern without inflating its significance. It placed the decision-making authority where it properly belonged, with the case officer on the ground. That the case officer had been recalled nine days later, before he could act on the recommendation, was not a consequence Foyle could have foreseen or prevented.

He made a note in his pad: Kirschwasser’s letter of 3 April — consistent with prior pattern of inflated threat perception (see October ‘73 and January ‘74 precedents). My response appropriate and proportionate per standing operational guidelines. The failure to action the contact protocol review is a direct consequence of Harbird’s recall, which occurred before Bonn Station could implement desk-level recommendations.

He underlined direct consequence and moved on.

There was a small brown spot on the next page of the file, and Foyle thought for a moment it was a tea stain — he had occasionally, in the early days of his career, permitted himself a cup at the desk, a habit he had abandoned around the time their son was born — but on inspection the spot was the paper itself, damaged from behind. His handwritten annotation on the reverse side, a marginal note in the iron gall ink they had all used before the Service switched to ballpoint in 1979, had oxidized over the decades. The iron compounds had attacked the cellulose fibres, eating through the page from the other side. He had noticed the same deterioration on several documents earlier in the file — his annotations from the mid-seventies were the worst affected. Someone in the Archives Section should have transferred these to acid-free folders years ago. He made a note of it in the margin of his pad.

The afternoon brought the file’s middle section: the operational assessments. These were the quarterly evaluations of DOVETAIL’s intelligence product, grading Kirschwasser’s material on the standard reliability scale from A1 — wholly reliable source, confirmed by independent means — down through the gradations of confidence to D4, which meant unassessed single-source material of unverifiable provenance. When DOVETAIL was initiated in March 1973, Kirschwasser had been assigned an initial rating of B2: usually reliable source, information probably true. The assessments were compiled by the responsible desk officer. By Foyle.

He found the critical document — the assessment from March 1974, the last before the recall. He had downgraded Kirschwasser from B2 to C3: fairly reliable source, information possibly true. The downgrade was supported by a three-page analysis of the asset’s recent product, written in Foyle’s measured prose, identifying inconsistencies in Kirschwasser’s reporting on Warsaw Pact military liaison protocols and noting that two of his claimed sub-sources within the East German signals directorate could not be independently verified. The analysis was thorough. The reasoning was sequential. Each observation led to the next with the logic of a man laying bricks. The conclusion — that Kirschwasser’s reliability had degraded to a level where his intelligence product should be treated with increased analytical caution — followed from the evidence as presented with the inevitability of water finding its level.

Foyle read his own analysis with the detached attention he would give to a colleague’s work submitted for peer review. It was, he decided, sound. Possibly conservative — if anything, a case could have been made for D1, which would have triggered a formal review of the asset’s continued utility — but sound. The inconsistencies were real. The unverifiable sub-sources were a genuine concern. The fact that the downgrade had subsequently been used by others, further up the decision chain, to justify a diminished response to Kirschwasser’s security warnings was how the system functioned. A C3 source reporting possible surveillance was an entirely different administrative proposition from a B2 source reporting the same. The Joint Intelligence Committee’s response matrix was explicit on this point: at B2, a surveillance report triggered an immediate operational pause and a full security investigation; at C3, it triggered only a review of tradecraft procedures at the local station’s discretion. The difference was a matter of classification. Foyle had provided the classification. He had not designed the response matrix. He had not authored the JIC guidelines. He had simply provided an honest professional assessment, and the system had processed it according to its own logic.

He closed the assessment folder and sat for a moment with his hands placed flat on either side of the open file, a posture he had adopted decades ago — sitting in judgment, as it were, over the accumulated paper — and which Vivian had once, in one of those rare moments of direct observation that had punctuated their marriage like flares over a dark landscape, called his “sentencing position.” Foyle was not aware of this name.

On Sunday he telephoned Vivian, as he did every Sunday at precisely eleven o’clock, ringing from the hall phone, which stood on a small walnut table with a drawer that stuck. The table had been Vivian’s. He had offered to send it to Salisbury with the rest of her things when she moved three years ago, and she had said it wasn’t worth the cost of the van. Most of her things, she had said, weren’t worth the cost of the van. The statement had seemed to Foyle at the time to be a comment about furniture.

She told him about the garden, which was coming along despite the frost, and about a concert series at the cathedral she was considering — Handel, she thought, or possibly Purcell, she could never keep them straight — and about a walk she had taken along the river path where she had noticed, on the side wall of a Norman church, an old Ordnance Survey benchmark: a small brass flush bracket set into the stone, marking a precise height above mean sea level.

“Do you remember those?” she said. “We used to spot them on our drives.”

“I remember,” Foyle said, and what he remembered, more than the benchmarks themselves, was the quality of those drives — the A-roads through Hampshire and Wiltshire on Saturday afternoons, Vivian reading the Ordnance Survey sheet in the passenger seat, calling out features and grid references, and Foyle absorbing the information without comment, navigating by her voice. They had been good at that.

“They’re part of a measurement system nobody uses anymore, apparently,” Vivian said. “GPS made them obsolete. But they’re still there. Set into the stone.”

“Yes,” Foyle said. “I imagine so.”

“How is the project?” she asked, because he had mentioned, in the vaguest possible terms consistent with the obligations of his pension, that he was reviewing some old work materials. She had received this information with the same courteous absence of interest she had brought to all his professional disclosures over forty years.

“Coming along,” he said. “A few discrepancies to work through.”

“Don’t exhaust yourself.”

“No.”

There was a pause — not uncomfortable, not pregnant, just the sound of an open telephone line between two people who had said everything they were going to say. Vivian mentioned that their daughter had rung about Easter arrangements. Foyle said he would check his calendar. They said goodbye. Neither said anything that a transcript would have identified as significant.

After he rang off he sat with his hand on the receiver for a moment longer than the action required. The hallway was cold — the radiator had been turned off as an economy — and the light through the frosted glass of the front door gave the narrow space a quality of permanent dusk. He looked at the walnut table, the stuck drawer, the small dish where he kept his keys. Vivian’s things. Or rather, the things Vivian had not considered worth the cost of moving. He removed his hand from the receiver and returned to the study.

The final section of the DOVETAIL file was the aftermath — the damage assessment compiled in October 1974, the internal reviews, the correspondence with liaison services in Bonn and Washington, the personnel actions and reassignments that followed any operational failure, though the word failure was never used in the documents themselves. The preferred term was operational conclusion at variance with projected outcomes, a piece of institutional language that Foyle had always considered more honest than it appeared, since it acknowledged that operations had projected outcomes at all.

He worked through it methodically, noting dates, cross-referencing names and positions against the timeline he had constructed on the folding table beneath the window. The light was failing. He switched on the desk lamp.

Kirschwasser had been arrested on the fourteenth of September 1974, five months after Harbird’s recall and three months after the last dead-drop communication. The arrest report, obtained by Bonn Station through a secondary source eighteen days later, indicated that the Stasi’s Hauptabteilung II — counter-intelligence — had possessed specific and detailed knowledge of the asset’s contact schedule, including meeting times, primary and secondary locations, and the dead-drop protocols that Kirschwasser had used exclusively for communication with his British handler. This information could only have originated from Bonn Station’s internal operational records, which were maintained in the station registry and copied, per standing distribution requirements, to the Europe Controllerate desk in London. To Foyle’s desk.

Foyle noted this without particular emotion. The security breach, wherever it had occurred, was a systemic failure — a failure of protocols, of compartmentalization, of the physical security arrangements at either Bonn Station or Century House. He had argued precisely this in a memo to Innes dated November 1974, which was included in the file and which he now reread with the satisfaction of finding his position vindicated by time. Systemic failures required systemic explanations. To blame an individual was to misunderstand how an intelligence service operated. The service was a machine. Machines malfunctioned. You repaired the machine. You did not prosecute the gear.

He had nearly finished. The damage assessment was followed by the personnel file extracts — Harbird’s reassignment to a training role at Fort Monckton, his early retirement in 1981, the absence of a forwarding address or pension correspondence beyond 1985. Harbird had simply vanished into the civilian world, as so many of them did. Foyle made a note to look into what had become of him, though the note had the quality of something written to complete a checklist rather than to initiate an action.

It was at the very bottom of the file — tucked behind the damage assessment’s final appendix, as if someone had added it later, slipping it into the folder the way you might slip a banknote into a book you were returning to a library — that Foyle found the supplementary memorandum.

It was typed on heavy cream paper, the kind reserved for communications above a certain classification grade. Dated June 1978. Bearing the handling restriction STRAP 3 — a level of compartmentalization well above Foyle’s clearance in 1974, in 1978, and for that matter at any point in his career. He should not have been reading it.

The memo was authored by Margaret Lowe, who had succeeded Innes as Head of the Europe Controllerate in 1976. Foyle had known Lowe slightly — a precise, humourless woman who wore her authority the way other women of her generation wore pearls: as a settled fact that did not require discussion. The memo was addressed to the Deputy Chief, Personal, and ran to two pages, single-spaced. The subject line read: DOVETAIL — OPERATIONAL RETROSPECTIVE AND STRATEGIC ASSESSMENT.

Foyle read:

The DOVETAIL operation achieved its primary objective, namely the reinforcement of DOVETAIL-3’s bona fides with the opposing service through the controlled release of corroborating intelligence material via DOVETAIL-7’s established reporting line. The operational costs, including the loss of DOVETAIL-7, were anticipated and accepted at the planning stage by the authorizing committee. The Committee’s assessment, which I endorse, is that the investment yield was proportionate to the expenditure and that DOVETAIL-3’s enhanced position within the target service provides continuing strategic value in excess of the original projections.

He read it again. He removed his glasses, cleaned them on the cloth he kept in the desk drawer, replaced them, and read it a third time, slowly, parsing each clause.

Lowe was claiming that the operation had succeeded. That the loss of Kirschwasser — the arrest, the interrogation, the presumed imprisonment or worse — had been anticipated. Planned for. That there existed a DOVETAIL-3, a separate and apparently more valuable asset within the East German intelligence apparatus, whose credibility with his own service had been enhanced by the controlled sacrifice of Kirschwasser’s intelligence product. In other words: Kirschwasser had not been compromised by a security failure. He had been fed to the opposition. Deliberately. As currency.

Foyle set down the memo. He removed his reading glasses again and placed them on the desk to the right of the notepad. He picked up his pen. He put it down. He picked it up again.

Lowe’s retrospective assessment (June 1978) contradicts the operational summary and damage assessment at several material points, he wrote. The claim that DOVETAIL-7’s loss was “anticipated and accepted at the planning stage” is not supported by any documentation in the operational file and is flatly inconsistent with the damage assessment of October 1974, which attributes the loss to a security breach of unknown origin. Lowe’s memo appears to be a retrospective justification, possibly motivated by the need to secure continued funding and authorization for DOVETAIL-3 (identity and status unknown to this reviewer). The absence of any reference to DOVETAIL-3 in the operational file prior to this memo is itself suggestive of post-hoc fabrication. Recommend the Histories Section examine Lowe’s claims in the full context of the 1978 budget cycle and her own position vis-à-vis the Controllerate’s performance review.

He underlined post-hoc fabrication and felt the small, particular satisfaction of having found the right phrase. That was what it was. A successor inheriting the wreckage of a failed operation and, rather than accounting for it honestly, claiming credit for the destruction. Transforming incompetence into strategy. It was the sort of thing that happened in any large institution — the Army did it, the Foreign Office did it — but Foyle had always believed the Service was better than that, or at least more careful about it. To find Lowe engaged in so transparent a rewriting of the record was both disappointing and clarifying. It explained why the file had remained classified for so long. Not to protect sources or methods. To protect a lie.

None of this occurred to Foyle in the way that ground occurs to a man who has always stood on it: that if Lowe’s memo was accurate, then the routing slip bearing his signature was not a clerical anomaly but a mechanism.

By five o’clock the light outside had gone entirely. The street lamp at the end of the drive had come on, casting an orange band across the study ceiling that Foyle had long since ceased to notice. He reviewed his notes from the day. He was satisfied. The file, taken as a whole and assessed with the dispassion that half a century of intelligence work had trained into him like a second pulse, confirmed his reconstruction: DOVETAIL’s failure was attributable to Innes’s premature recall of the case officer, compounded by systemic security failures at Bonn Station. The Lowe memorandum was an aberration — a retrospective confection that did not survive comparison with the contemporaneous record.

He wrote his conclusions in the ruled pad, in the same careful hand that had initialled the routing slips and the assessments and the operational correspondence thirty years and more before:

DOVETAIL — preliminary findings. The operational failure is attributable primarily to the premature withdrawal of the Bonn case officer (see routing analysis, pp. 34-51) and secondarily to inadequate security protocols regarding DOVETAIL-7’s contact schedule (see damage assessment, pp. 78-94). The retrospective assessment by M. Lowe (1978) is inconsistent with the operational record and should be regarded with appropriate scepticism. Recommend the Histories Section re-examine the role of G. Innes in the decision to recall Harbird, with particular attention to the absence of minuted authorization for the Personnel Movement order of 12 April.

He capped his pen — a Parker, the same model he had used at the Controllerate, though he’d replaced the nib twice and the barrel once, which raised a question about identity he had never thought to ask. He folded his reading glasses into their leather case and placed the case in the top drawer of the desk. The notepad he left open so the ink could dry, though the Parker used a modern cartridge and the precaution was unnecessary. He did it anyway.

In the kitchen he filled the kettle and stood at the window while it boiled. The neighbour’s garden backed onto his, separated by a fence whose panels had begun to warp and bow. One panel had come loose entirely in the January gales and now leaned against its neighbour at an angle that suggested resignation rather than collapse. Foyle had spoken to the neighbour about sharing the cost of replacement. The neighbour had said he would get a quote. That had been six weeks ago.

He made tea — Earl Grey, no milk, which was what he drank now. Vivian had preferred builders’ tea, strong, the colour of creosote, with enough milk to cool it for immediate drinking. He had accommodated this preference for the duration of their marriage without once asking himself whether he shared it. When she moved to Salisbury, he had gone through the kitchen cupboards with the same methodical attention he brought to a file review, and it had taken him nearly a week to discover that he preferred Earl Grey.

He carried the cup back to the study and stood at the window. The orange light made everything the same colour — the pavement, the parked cars, the bare branches of the flowering cherry that Vivian had planted their first spring in the house and that Foyle had maintained since with the conscientious regularity he brought to everything. A woman was walking a small dog on the opposite pavement. She was wearing a coat too thin for the weather and the dog was pulling at its lead with the concentrated determination of an animal that knows exactly where it wants to go.

Foyle drank his tea. It was slightly too hot. He held the cup with both hands, which was something he had started doing in the past year. Circulation. His GP had mentioned it.

He returned to the desk and began closing the file. Each document was replaced in its original position, secured with the stainless-steel archive clips he ordered from a specialist supplier in Holborn who catered to solicitors and museums. The folder went back into the banker’s box, which he carried to the shelving unit and placed in its correct position between 1973 and 1975. When he lifted the lid to settle the folder properly, he noticed that one corner of the cardboard had gone soft. Condensation. The unheated room. Moisture had accumulated on the metal shelving overnight and wicked up into the box. The label on the end — DOVETAIL — OPERATIONAL 1974, in his own hand — had begun to bleed where the damp had reached it. The ink was spreading into the softened fibres, the letters losing their edges, the careful handwriting dissolving into something that was still legible but would not be for long.

He would need to relabel it. He had a roll of adhesive labels in the desk drawer, the same stock he used for all the boxes, and the job would take two minutes. He would do it tomorrow. He pressed the lid on, each corner firm, and pushed the box flush against the others.

Foyle switched off the desk lamp. Tomorrow he would type his conclusions and post them to an address in Whitehall that might or might not still accept unsolicited correspondence from retired officers. If it did not, he would try another route. There were always routes.

He closed the door behind him and went upstairs to wash his hands before supper.