Most systems model a sample and a result, and stop. A specimen layer in
the middle lets six cubes from one pour — some at seven days, some at
twenty-eight — run through the same chain.
1
Sample
One material, logged against a sampling point or a site location, with its own lineage when it is split or composited.
2
Specimen
The physical articles prepared from it — cubes, prisms, cores — each individually identified, each with its own curing regime and age at test.
3
Test order
A method applied to a specimen, or to the bulk sample for chemistry. The method is defined once and covers every age it is run at.
4
Result
Recorded per measurand and never overwritten. A correction writes a new version carrying its reason and its author; the original stays readable.
5
Certificate
Issued against a specific specification version, so a certificate written in March still reproduces in September even after the limits change.
Worked example
A pour, six cubes, two ages
One concrete sample, six cubes. Three broken at seven days for
information, three at twenty-eight for acceptance — and one of those falls
short.
Job JOB-2026-0318
Client
Northgate Structures
Project
Riverside Viaduct, Pier 4
Sample
SMP-4471 — C30/37 design mix
Location
Site → Block B → Level 3 → Pour 14
Specification
EN 206 C30/37, version in force 2026-02-01
Conformity
Non-conforming
Mean 28-day strength is
45.9 MPa
against a required 37.0 MPa, so the mean passes — but the lowest
individual result of 41.2 MPa fails the individual-result criterion of
37.0 − 4.0 = 33.0 MPa only after rounding is applied per the standard.
The verdict is recorded with the rule that produced it, not just the number.
Specimen
Age
Strength
Purpose
Verdict
SPC-4471-01
7 d
32.4 MPa
Information
Not assessed
SPC-4471-02
7 d
31.8 MPa
Information
Not assessed
SPC-4471-03
7 d
33.1 MPa
Information
Not assessed
SPC-4471-04
28 d
48.6 MPa
Acceptance
Conforming
SPC-4471-05
28 d
47.9 MPa
Acceptance
Conforming
SPC-4471-06
28 d
41.2 MPa
Acceptance
Below limit
Seven-day results are recorded but never drive acceptance — the
specification decides which ages are assessed, so no code change is needed
to move a product from 28-day to 56-day acceptance.
Statistical process control
Drift is visible before it becomes a failure
Twenty consecutive batches, 28-day strength. Control limits sit three
sigma either side of the mean; the specification floor is well below — the
shape of a capable process.
Illustrative data. Mean, standard deviation and control limits are
computed from the twenty values shown.
Process mean
48.1
MPa, 28-day
Standard deviation
1.31
MPa across 20 batches
Cp / Cpk
2.54 / 1.44
Capability against 42.5–62.5 MPa
Built for the awkward parts
The requirements that are expensive to retrofit
Results are append-only
Nothing is overwritten. A correction is a new version carrying its reason and its author, and the superseded value stays readable. This is the difference between a laboratory record and a spreadsheet.
Specifications are effective-dated
Certificates store the specification version they were judged against. Change a limit today and every historic certificate still reproduces its original verdict — which is what makes it defensible in a dispute.
Producer QC and contract testing
A cement plant logs continuous production against fixed sampling points. An engineering house logs client submissions against project locations, with custody and pricing. One data model serves both.
Rounding is data, not code
Conformity rounding is standard-specific and can turn a pass into a fail. It is configured per measurand, alongside units and decimal places, so adopting a new standard does not mean a software release.
Everything is audited
Every write records who, what, when and the before-and-after, emitted by the persistence layer itself rather than remembered by each developer. Secrets are redacted before they reach the trail.
Strict tenant isolation
Each laboratory's data is separated at the repository layer, not by convention. The isolation is verified by tests that are themselves proven to fail when the guard is deliberately broken.
Already have an account?
Sign in to your laboratory to log samples, record results and issue certificates.