Why do these two charts differ?

Two calculations of the same birth chart can disagree. Usually the reason is written down in the records themselves — a different house system, a different moment, a different convention — and reading it off is faster than arguing about it. This page does that reading, in your browser.

It reads the calculation records this site produces: download one from abirth chart, let thedeveloper starter write one, or ask thelocal MCP server for output: "record"and keep the record field of its reply — that field is the record, and the reply around it is not one. They arezodiacs.natal-envelope.draft-v1, a Zodiacs-owned draft vocabulary rather than an industry interchange format — so another program's export and a screenshot are both out of scope, and the four examples below are the fastest way to see what an answer looks like.

Differences come first, because they are facts: both files are read, and every value that disagrees is listed — inputs, conventions, angles, house cusps, every field of every body, and the aspect list. Longitudes are compared around the circle, so 359° and 1° are two degrees apart rather than 358. Two numbers that print the same at the six decimals these records carry are called a rounding difference rather than a different calculation; the test is what they print, not a tolerance, because rounding is a step and no tolerance sits exactly on it.

Explanations are then labelled by the evidence behind them:

• Reproduced — recalculated on this device, changing one setting and leaving everything else alone, and the result matched.
• Reported — stated by the files themselves, with nothing to compute.
• Hypothesis — fits the evidence, not demonstrated. Several hypotheses can fit one difference, and all of them are listed.
• Unresolved — nothing in either file accounts for it.

A recalculation only ever runs on engine 0.1.1-rc.6, the version this page carries. If a record was produced by a different version it is not re-run and presented as the original: the difference stays a hypothesis and the limit is stated. A version, checksum or source URL inside an imported file is a claim that file makes about itself, not proof that it is genuine.

Reproduced is withheld unless the recalculation demonstrated the values that actually moved. Two polar charts that both fell back to whole sign have the same cusps however they were requested, and matching values neither file disputes demonstrates nothing. Only the house system is re-run: a different moment or place stays a hypothesis. Rows left over once each candidate is matched to what it can account for are listed as unresolved, by name.

Both sides of a comparison usually come from the same engine. Agreement between two runs of one engine is consistency, not an independent check of astronomical accuracy.

Supported: the calculation record this site offers for download from a birth chart, and the records the developer starter produces. Both arezodiacs.natal-envelope.draft-v1, validated with the engine's own parser and its size, depth and node limits. A record from a different engine version of that same schema is read, and the version difference is reported.

Not supported, and not silently reinterpreted as something else: charts from other software in their own formats, PDFs, images and screenshots. There is no claim of general third-party compatibility here.

Your records never leave the device. The files are read in the browser, compared in memory and dropped when you clear them or close the page. No account, no upload, no birth details in the address bar, and nothing written to storage. The optional summary download leaves out dates, coordinates and positions but keeps the exact difference between them, so anyone holding one of the two charts can recover the other: safer to share than a full record, not anonymous.