This report documents the Institution's first direct evaluation and attempted containment of the entity provisionally designated "Achemon," linked to an active infernal contract with the subject of Case 4471-H. The objective was threefold: confirm the entity's existence and nature through direct observation, determine its true classification tier, and if possible, achieve containment via standard binding protocols.

As of filing, the evaluation team is unable to confirm success on any of these three objectives with confidence. The reasons for this are detailed below and constitute, in the opinion of the lead evaluator, a significant institutional concern.

Initial scrying attempts conducted on 17–19 March by the Perception Unit returned no usable data. Standard clairvoyance protocols (Weir Method, Aldiss Sweep, Kessler Resonance Mapping) each returned null results — not failure states, but null, as though there were nothing to find. The entity does not register on any spectral band currently in institutional use.

More troublingly, members of the Perception Unit reported cognitive effects during the scrying sessions that are difficult to attribute to fatigue or environmental factors. Operative Sandoval reported losing approximately forty minutes during a session on the 18th with no memory of the interval and no recording anomalies to corroborate the gap. Operative Chen noted that during her session she became gradually uncertain whether she was searching for an entity or whether she had invented the concept of searching for an entity. She describes the experience as "an erosion of the premise," and requested to be removed from the assignment.

It should be noted that this pattern — an entity that is not invisible but rather unconvincing — has no close analogue in the Institution's taxonomy. The nearest precedent is the Holford Iteration (1991), though the mechanisms appear dissimilar. A request has been filed to re-examine the Holford case notes in light of these findings.

Cross-referencing the client's testimony with historical infernal behaviour models, Dr. Achterberg proposed a manifestation hypothesis: entities of this nature, which resist direct observation, may become briefly perceptible in environments saturated with acute collective suffering — sites where misery and desperation are concentrated to a degree that warps local thaumic conditions. The theory is that such environments force the entity into greater coherence. It feeds, or at minimum gathers, and in doing so becomes momentarily visible.

The evaluation team began monitoring for a suitable observation site. On 22 March, the ████████████████ industrial fire in ██████████ provided the necessary conditions: significant casualties, sustained public attention, prolonged emergency response, and a concentrated zone of grief within a defined perimeter. The Institution's Ethics Board granted a limited observation window of six hours, conditioned on non-interference with emergency operations.

Human observers were deployed to the perimeter of the disaster site at 14:20 on 22 March. Within minutes of arrival, the forward team reported the same cognitive difficulties noted by the Perception Unit — a sense that their purpose at the site was unclear, that they had perhaps arrived at the wrong location, that the operation had already been cancelled. Morrow, who had been briefed on these effects and was wearing a Voss-Kessler cognitive filter, was the only member of the team to maintain consistent operational focus. He described the sensation as "like trying to remember a word that keeps dissolving."

Given the inadequacy of direct human observation, the team fell back on modified drone surveillance. The drones (four units, standard commercial quadcopters) had been fitted with Aldiss-spectrum camera filters and a thaumic resonance overlay developed by the Instrumentation Lab. The modification was experimental and had not been field-tested against an infernal-class entity.

The drones were deployed at 15:07 and directed toward the still-burning wreckage at the centre of the site.

For approximately ninety seconds after signal loss, the team assumed equipment failure. Then Drone 1 reappeared on radar. It had reversed course and was approaching the forward team's position at speed.

Within seconds, all four drones were airborne again and moving in coordinated patterns that bore no relationship to their programmed flight paths. The control units were unresponsive. The drones were operating autonomously — or rather, they were operating under the direction of something that was not the team.

The drones harassed the forward team for approximately four minutes. "Harassed" is the word used in the field report, but Morrow's audio log paints a more specific picture: the drones were circling individual team members, descending to head height, then pulling away — not attacking, but herding. Morrow describes them as behaving "like sheepdogs working a flock toward a gate, except there was no gate and we couldn't see the shepherd."

At one point Drone 3 hovered in front of Operative Yilmaz's face for eleven seconds. Yilmaz later reported that the camera light was on and that she had the distinct impression she was being studied.

At 15:24, Morrow made the decision to fire the Ackroyd Mark VI containment device (the "god trap") at the area of highest thaumic concentration as indicated by the last reliable drone reading. The device was aimed at the northwest corner of the wreckage and discharged at 15:24:33.

Upon discharge, all four drones fell from the sky simultaneously, as though their strings had been cut. They struck the ground within a two-second window. Post-recovery analysis shows that each drone's firmware had been wiped to factory default. The Aldiss-spectrum modifications were intact but the calibration data had been zeroed. The Instrumentation Lab has described this as "not something a signal disruption does. This is something that understood the modification and removed it."

The Ackroyd Mark VI was recovered and transported to the Institution's secure evaluation facility under full containment protocol. Upon opening the trap's observation port, the containment team found a single common pigeon (Columba livia domestica), alive, apparently uninjured, and displaying no behavioural abnormalities.

The pigeon was transferred to Secure Biological Containment Unit 7 (the same unit used for quarantine of potentially compromised organisms after the ██████ event in 2019). A full evaluation battery was initiated immediately.

Every test available to the Institution indicates that the contents of the Ackroyd Mark VI are, in fact, an ordinary pigeon. There is no thaumic signature. There is no spectral anomaly. There is no behavioural irregularity. The animal appears to be approximately two years old, in good health, and mildly underweight — consistent with a feral urban pigeon. A leg band indicates it may be a lost racing pigeon, registration pending verification.

The defining characteristic of this entity is that it resists not just observation but the certainty of observation. Its primary defence is to make the observer unsure whether they have perceived anything at all. A pigeon is the perfect expression of this capability. There is no result more disarming, more discrediting, more thoroughly mundane than opening a god trap and finding a pigeon inside. It is the kind of result that makes the evaluator question whether the operation was ever warranted.

Which is, of course, exactly what this entity does to people.

Given the entity's demonstrated ability to commandeer autonomous systems, the following protocols have been implemented for Containment Unit 7:

All automated systems within the unit — environmental controls, surveillance feeds, door actuators — have been replaced with manual equivalents. Lighting is provided by caged incandescent bulbs on a physical switch. Temperature is regulated by a manually adjusted ventilation damper. The surveillance camera has been removed; observation is conducted through a reinforced glass viewport by a human observer on rotating four-hour shifts.

Feeding and maintenance of the pigeon is conducted once daily by a designated handler in full Level 3 protective equipment (thaumic-shielded suit, Voss-Kessler cognitive filter, iron-weave gloves). The handler enters alone, provides seed and fresh water, removes waste, and exits. Duration inside the unit is capped at eight minutes. The handler is debriefed by a cognitive integrity officer immediately after each session.

The evaluation team's position can be summarised as follows:

There is a significant probability — estimated by consensus at 65–70% — that the entity designated "Achemon" remains at large and that the Ackroyd Mark VI captured an ordinary pigeon through coincidence of timing. The capture attempt may have succeeded only in alerting the entity to the Institution's involvement and capabilities.

There is a smaller but non-negligible probability — estimated at 25–30% — that the pigeon is the entity, contained in a form so thoroughly mundane that our detection methods cannot penetrate the disguise. If this is the case, the containment is holding, and the entity's apparent docility may indicate that the Ackroyd device is functioning as intended.

There is a residual probability — less than 5%, but formally acknowledged — that the entity was captured and has since escaped the trap by some means, leaving the pigeon behind as a distraction. This scenario is considered unlikely but cannot be excluded.