RESEARCH DIRECTORATEPX–TB–026
REVISION 1 / PUBLIC EDITION
PARALLAX / Technical brief
The observer gap
Six instruments returned a complete account. The interval between them did not.
01 / Observation
In the P-09 chamber series, six independently trained observers monitored a repeating material transition. Each returned a continuous event record. A separate timing witness registered a 240-millisecond interval that none of the observers represented. No instrument reported a fault.
02 / Method
The timing witness did not observe the transition itself. It recorded the latency between external stimulus, internal acknowledgment, and publication of each observation. A common discontinuity in that sequence identified an omission without supplying the omitted content. Removing any single observer did not eliminate the discontinuity.
03 / Result
The omission was detected in 48 of 48 controlled trials with an injected common-mode masking condition. In 48 unmasked controls, the detector reported no gap. These counts describe a bounded chamber result; they do not establish a field error rate. All six primary observers continued to score as internally consistent.
04 / Interpretation
Consistency checks operating only on a published record cannot detect every process that made the record incomplete. A system must retain evidence about the production of observations, not merely the observations themselves. Agreement is useful. It is not an independent witness.
05 / Boundary
The detector establishes that an interval is unaccounted for. It does not identify what happened within it. The chamber geometry, masking apparatus, and witness implementation are withheld under D-2. No reconstructed account of the missing interval is authorized by this brief.