Both flush values were recoverable for all 2,778 accepted dual-flush identities
In the EPA WaterSense directory snapshot saved on October 30, 2026, our accepted toilet dataset contains 2,778 dual-flush identities. This audit recovered a full-flush value and a reduced-flush value for every one of them, with no missing mode value in that accepted subset. The pairs form 50 distinct numerical combinations. This is a result about a saved dataset, not a claim that every toilet currently on sale has complete documentation.
The practical lesson is to preserve the two values together. A directory search that returns only one number may have lost information during extraction or display. Before filling a blank with zero, check the original cell and confirm that the search is looking at the same product identity and snapshot.
Define the population before counting blanks
Our source archive contains a tank-type toilet CSV with 6,793 data rows, including 2,919 rows marked Dual and 3,874 marked Single. After the previously documented identity, duplicate and quarantine processing, the research dataset contains 6,591 accepted identities: 2,778 dual and 3,813 source-scalar records. An accepted identity is the dataset's composite product key; it is not a sale, a household or a popularity vote.
| Population | Count | Appropriate interpretation |
|---|---|---|
| Original toilet CSV rows | 6,793 | Rows before identity filtering |
| Original rows marked Dual | 2,919 | Raw records, potentially repeated or excluded |
| Accepted toilet identities | 6,591 | Population used in our toilet research |
| Accepted dual-flush identities | 2,778 | Denominator for this pair-completeness audit |
The difference between 2,919 raw dual rows and 2,778 accepted dual identities is 141. It would be wrong to call those 141 “toilets missing a reduced flush value.” The subtraction crosses two processing stages. It does not identify a missing-value category. Establish the reason for excluding each source row before assigning any cause to the difference.
How we checked the original records
We verified the saved archive's SHA-256 fingerprint, read all accepted toilet flush-volume fields across the normalized partitions, and checked each field's row locator against the original CSV cell. All 6,591 accepted fields matched their saved raw text. We then independently parsed the full and reduced numbers for all 2,778 dual records and compared them with the normalized values using decimal arithmetic. Every pair matched; none required a zero, average or guessed replacement.
The source archive fingerprint is 577fa416addade9df0ee69c78d9e7c46ac4098b933286d2515e1bb18f8ca78ef. Row locators are relative to that archive, including its header row. They should not be transferred to a newer download whose ordering may differ. The audit does not independently certify the underlying products; it checks whether our saved representation faithfully preserves the source values.
The three most common pairs illustrate the denominator problem
| Full / reduced, US gallons per flush | Accepted identities | Share of 2,778 dual identities |
|---|---|---|
| 1.60 / 1.10 | 1,238 | 44.56% |
| 1.28 / 0.80 | 320 | 11.52% |
| 1.60 / 0.80 | 306 | 11.02% |
Together these three combinations account for 1,864 identities, or 67.10% of the accepted dual-flush subset. They are common in this directory snapshot, not necessarily common in installed bathrooms. A manufacturer with many separately recorded model identities can contribute many entries without establishing any sales share.
The first pair's 1,238 identities are 44.56% of the dual subset but only 18.78% of all 6,591 accepted toilet identities. Both divisions are mathematically valid; they answer different questions. Name the population next to every percentage, especially when comparing an all-toilet chart with a dual-only table.
What a complete pair still cannot answer
Two populated fields do not establish the frequency with which a household uses each mode, actual installed water consumption, physical compatibility or current certification. Nor does a smaller reduced-mode value by itself establish a better overall choice. To estimate a household total, retain both values and supply explicit event counts; see our dual-flush use-scenario analysis. That calculation is separate from this completeness audit.
For a current buying decision, consult EPA's current product search and the exact manufacturer's documentation. The live directory may differ from our saved population. Use our toilet research page for the historical distribution and methods for source and identity limits. If a current listing has an unexplained blank, record it as unknown rather than borrowing a number from this historical aggregate.