Production QA Log 2014–2017: Four Years of Batch Records
This is the working quality log kept during contract production of the flagship golden lager. Each row is one sample at one stage of one brew: turbidity at two spin levels, pH, color, bitterness, gravities, real and apparent extract, RDF, pounds of extract, calories — and, on twenty-one rows, the words "Gliadin Negative."
The dataset
155 records, 2014–2017, full precision in the CSV. A representative brew (BBGL #5, June–July 2014) reads like this:
| Date | Stage | FTU 90 | pH | OG °P | ABV % | RDF % | Note |
|---|---|---|---|---|---|---|---|
| 06-04 | W | — | 5.52 | 16.1 | — | — | |
| 06-16 | R | — | 3.62 | 16.2 | 7.02 | 66.1 | Gliadin Negative |
| 06-30 | C | 272 | 3.51 | 16.1 | 6.96 | 65.9 | |
| 06-30 | F | 226 | 3.71 | 11.3 | 4.73 | 65.1 | Gliadin Negative |
| 07-03 | X | 124 | 3.57 | 11.2 | 4.68 | 65.0 | Gliadin Negative |
Reading notes
- The gluten testing is in the batch record, not the marketing. Twenty-one rows carry "Gliadin Negative," recorded at fermentation, filtration, and final stages across the years — the trust chain, practiced as routine.
- The high-gravity dilution is visible in the data. OG sits near 16.5°P through the early stages, then drops to ~11°P between the C and F stages — the documented extract-era practice of fermenting strong and cutting to package strength, showing up in raw numbers years before the procedure document that describes it.
- Turbidity tells the clarity story. FTU values run to the thousands mid-process and land in the tens at the final stage — the practical wort-clarity battle this site's whole evaluation argument turns on, measured batch by batch.
- RDF holds 65–68% for four years — the consistency claim behind the published spec (RDF 68 ± 1), demonstrated rather than asserted.
- Stage letters are as recorded. W, R, C, F, X were the log's own shorthand and the file does not define them. Reading the gravities: W is knockout wort (no alcohol yet) and X is the final packaged state — that much the numbers say; the rest is left as recorded rather than guessed.
- One column is unlabeled in the source and is preserved as
unlabeled_value, unexplained, because inventing a label would be worse than admitting one is missing.
What was removed
The facility identifier column. This log was kept at a contract production facility; the facility is not named on this site, and its single-code column is the only thing withheld from the record. Tank codes, sample IDs, and every measurement are as written.
Download the dataset
→ Download as CSV — all 155 records, all columns, full precision.
Record
First published 2026-08-10 as v1.0. Corrections to date: none. This dataset is a frozen record — it changes only for corrections, and every correction is logged in this section with its date and what changed.
Where else this data lives: nowhere public — working QA logs never leave the brewery. This page is the exception, on purpose.
How to cite this dataset
Production QA Log, flagship sorghum beer, 2014–2017. Bard's production records, republished by Gluten Free Brewer, 2026. https://glutenfreebrewer.com/research/production-data/production-qa-log-2014-2017
Related
- LCBO Finished-Beer Certificates — the regulator's independent numbers on the same beer
- Building Bard's — the extract era — the process these batches ran
- Gluten Testing Methods and Their Limits — what a gliadin test does and does not prove
Source and Validation Notes
Transcribed programmatically from the original workbook (two sheets, data spanning 2014–2017) using cell positions, not the text extraction — a column-drift quirk between sheets (the stage letter migrating between two columns) was resolved in parsing and both variants verified. Values are as recorded by the facility's QA bench; nothing has been recomputed. The gliadin notations record the log-keeper's entry, not a certificate — what that distinction means.