Mercer, as chief deputy United States marshal assigned to courthouse security, was the right first channel. He had already secured the scene and the recordings. He now opened a formal referral, in writing, documenting the time, the place, the persons involved, the steps already taken to preserve evidence, and the fact that the matter involved a federal judge and officers scheduled to testify in a pending case. That referral went, as such referrals must, to the Marshals Service through its chain, and to the appropriate oversight authorities in general terms, including the offices responsible for reviewing alleged misconduct by federal officers and for assessing whether any criminal referral was warranted. I deliberately did not name, in my own statement, every office that might ultimately touch the matter, because I did not want my statement to read as an instruction about jurisdiction. I stated the facts, I identified the evidence I knew of, and I left the question of which authority would lead to the authorities themselves, through their ordinary intake process. That is how oversight retains its independence: the complainant does not choose the investigator.
My formal statement was taken later that day, not in the corridor, not on the bench, but in a neutral interview room, by investigators who were not part of my chambers staff and not part of the courthouse security team that had responded first. I gave it in chronological order, without embellishment and without speculation about motive. I described where I had been, why I had been in the lower level, what was said, what was done, and what happened next. Where I was uncertain about a detail, I said so. Where I remembered a sequence but not exact wording, I said so. A statement that pretends to perfect recall is less credible than a statement that marks its own limits, and I wanted this statement to be credible above all else. I reviewed it, I corrected two small errors in it about times, and I signed it. A copy was retained by the investigators. I did not keep a personal copy to circulate. The statement belonged to the investigation, not to my narrative.
In parallel, the investigators began to identify other witnesses, and this is where the record began to do work that fear had prevented individuals from doing alone. Courthouses are hierarchical places. Staff members, junior deputies, clerks, and maintenance workers all understand, often without being told, that speaking about an incident involving officers and a judge carries risk. They worry about retaliation, about being disbelieved, about becoming known as the person who spoke. What changed for them that morning was not that someone gave them a speech about courage. It was that the court itself, on the record, had acknowledged that an incident had occurred, that it had been referred, and that evidence was being preserved. Once that acknowledgment existed, a witness did not have to be the first person to say that something had happened. The record had already said it. The witness only had to add what he or she had seen or heard, which is a smaller and safer step, and a more truthful one, because it does not require the witness to carry the whole case alone.
Camille Ross was the clearest example. When investigators spoke to her, she was, by every account I later received through proper channels, frightened at first, and careful, and reluctant to overstate what she knew. She was a clerk's office staff member who had been working in the records office behind the frosted-glass door, doing ordinary work, when the corridor sounds reached her. She had not gone looking for an incident. She had not positioned herself to observe one. She had simply been present, behind a door that muffled but did not erase sound, and on a telephone line that recorded what it captured. When she was asked what she had heard, she did not offer a theory. She described sounds, voices, timing, and the fact that she had remained at her desk because she did not know whether opening the door would help or make matters worse. That hesitation, which she described honestly, was itself credible. People who invent stories do not usually invent their own uncertainty.
Her telephone recording corroborated her account in the way only an objective record can. Investigators did not ask her to interpret it for them. They obtained it through the clerk's office, verified its metadata, confirmed the line and the time, and preserved it alongside the surveillance footage. The two records together, the audio from the records office line and the video from the corridor cameras, created a timeline that did not depend on any single person's memory. That timeline showed the proceeding upstairs going forward on time that same morning, showed the separation of the officers, showed the preservation steps, and showed, in the background, the ordinary work of the courthouse continuing around an extraordinary event. I did not review those recordings myself during the investigation. I had no need to, and my review would have added nothing except a question about influence. I trusted the process to do what it was designed to do, which was to assemble facts before conclusions.