Developer tools · Unix timestamp converter
Negative Unix timestamps: how dates before 1970 are represented
· How it works
timestamps unix-time data-formats
Unix time counts forwards from 1970, but it counts backwards too. This post explains what a negative epoch means, how −1 becomes the last second of 1969, and where systems that assume positive values break.
A date of birth with a minus sign — a user born in 1965 stored as a negative integer, and the code that treated it as an error
A record for a person born before 1970 can legitimately contain a negative Unix timestamp. Rejecting every leading minus sign converts a representation choice into a data-loss bug. ToolAcre’s parser accepts an optional sign, and `fromEpoch` marks an interpreted millisecond value below zero as `beforeEpoch` rather than treating it as invalid.
The display then adds a note explaining that the instant precedes the Unix epoch. This is stronger than silently rendering a date because it keeps the unusual sign visible to the reviewer. If an upstream schema forbids negatives, the failure belongs to that schema or migration, not to the arithmetic that maps counts on both sides of zero.
Signed integers and the epoch — why time_t is signed, and what −1 and −86400 correspond to
Zero is 1970-01-01T00:00:00.000Z. The repository tests prove that −1 second is the immediately preceding instant, `1969-12-31T23:59:59.000Z`. Likewise, −86,400 seconds is one fixed POSIX day before zero. Signed arithmetic makes the axis continuous instead of requiring a second epoch for older records.
The outline attributed signed storage to `time_t`, but this code does not establish the design history or type used by every platform. Its narrower evidence is enough: JavaScript numbers can be negative, detection uses absolute magnitude, and Date can format the result. Describe the producing field’s declared type separately from the converter’s accepted input.
Signed epoch arithmetic around zero, without claiming why every time_t chose its type
Negative values retain the same unit distinction as positive ones. `−1000` seconds points 1,000 seconds before the epoch, whereas `−1000` milliseconds points only one second before it. Automatic detection treats both as seconds because their magnitude is below 10¹¹, so a short pre-epoch millisecond count requires the explicit milliseconds selection.
This is a concrete failure of digit-count folklore. A developer seeing a minus sign and four digits cannot infer the unit from appearance. Use the source contract, choose the selector and inspect the unit note. ToolAcre’s magnitude rule is a default for typical data, not a replacement for metadata on historical values close to 1970.
Where negative values fail — systems that use −1 as a sentinel, unsigned storage, date pickers that clamp, and languages that reject them
Outside the converter, `−1` is sometimes reserved as a missing-value sentinel, and an unsigned column cannot preserve any negative instant. Those are common design possibilities, but the repository cannot certify how a particular database, date picker or language behaves. Test the actual boundary before moving archival dates through it.
A useful probe contains three values: null or the documented missing marker, −1, and a clearly older date. If −1 disappears while the older value survives, sentinel handling is implicated; if all negatives fail, inspect type and validation. Do not “fix” the symptom by adding 1970 years or taking an absolute value, because both create different instants.
Worked example: converting −14182940 — the arithmetic to a July 1969 date in UTC and in a local zone west of Greenwich
For −14,182,940 seconds, start at the epoch and move backward 164 complete 86,400-second days, leaving 13,340 seconds. The remainder is 3 hours, 42 minutes and 20 seconds. The resulting UTC reading is `1969-07-20T20:17:40.000Z`, an instant chosen here specifically for this article.
A local reading west of UTC may show an earlier wall-clock hour or even another date, but its exact result depends on the browser zone and Intl rules. Read it from the panel rather than publishing a universal answer. The fixed verification is the ISO row; the local row is an environmental annotation on that same negative count.
Worked example: converting −14,182,940 with verifiable UTC arithmetic
ToolAcre checks its interpreted value against ±8.64×10¹⁵ milliseconds, the Date range stated in source. It also proves conversion of −2,208,988,800 seconds to `1900-01-01T00:00:00.000Z`. Those facts show that the route reaches well before 1970 without implying that every storage system can carry the same range.
A signed 32-bit field has a narrower mathematical interval than JavaScript Date, but the converter does not inspect a field width when you paste a number. Preserve the source type in an audit. A successful browser conversion demonstrates representability here, not safe round-tripping through firmware, SQL columns or binary protocols elsewhere.
The converter’s Date range is known; external 32-bit limits depend on their storage type
Turning a count into a JavaScript Date does not reconstruct the civil calendar actually printed in every location centuries ago. The implementation uses the platform’s date model and returns an ISO representation; it contains no historical-calendar selector, locale archive or documentary source for calendar reforms. Those claims are intentionally absent.
For modern software records, the proleptic machine representation may still be exactly the contract you need. For genealogy, legal history or archival transcription, preserve the original written date and its calendar context alongside any derived epoch. A convenient sortable number should not erase uncertainty that conversion cannot resolve.
Calendar-history claims are outside this converter’s evidence
Negative is a direction on the epoch axis, not an error category. ToolAcre parses it, applies the same seconds-or-milliseconds rules by magnitude, and flags the result as before the epoch. That behaviour lets a reviewer distinguish a legitimate old instant from a sentinel or unsigned-storage failure elsewhere.
When auditing pre-1970 data, state the unit, source type and missing-value convention before interpreting the sign. Then compare the ISO result with a known record. The sign alone is meaningful arithmetic; the surrounding schema determines whether the producer intended that arithmetic or used the same number as a control value.