# C496 — Refinement defects

## Question

When do documented refinements preserve the reversed path value?

## Inputs

```json
{
  "MT_regular_Noah": [
    1650,
    [
      1050,
      600
    ]
  ],
  "MT_cumulative_Shem": [
    9170,
    [
      8570,
      600
    ]
  ],
  "MT_cumulative_Noah": [
    9170,
    [
      7620,
      1550
    ]
  ],
  "LXX_regular_Noah": [
    2250,
    [
      1650,
      600
    ]
  ]
}
```

## Sources

```json
[
  "File52c §3.3",
  "C489"
]
```

## Opened utc

2026-09-27T22:42:34.936653+00:00

## Results

```json
{
  "defects": {
    "MT_regular_Noah": 0,
    "MT_cumulative_Shem": 990,
    "MT_cumulative_Noah": 990,
    "LXX_regular_Noah": 990
  },
  "definition": "delta_refinement=sum I(parts)-I(sum parts)"
}
```

## Finding

The MT regular refinement has defect0; both cumulative refinements and the LXX regular refinement have defect990. Thus12026 versus13016 and10436 versus11426 are outcomes of the same path rule, not separate arbitrary anchors.

## Reassessment

Test whether appending the documented Conquest-to-Nativity leg gives a uniform backbone continuation.

## Checks

```json
{
  "raw_lengths_preserved": true,
  "defects": true
}
```

## Closed utc

2026-09-27T22:42:34.936928+00:00

## Record provenance

Exact record hash: `3bc4c41817c5a53cc1f35b997be3e74536789539edd1e7d6c9897faab3af971a`

Source journal: `Research_Cycles/C0482_C0531/journal.json`