My Two Production Scales Give Slightly Different Readings — How Do I Prevent Different Employees From Creating Different Formulations?

Why two scales can both be consistent and still disagree

Most teams notice the problem in a small way. Someone weighs a bottle on one scale, then puts it on the other to double-check, and the numbers do not match. Both scales return to zero, both give the same reading when you weigh the same object twice, and each employee trusts the one they use. The trouble is that repeatability and correctness are different properties. Each scale can be internally consistent while reading a different value from the other.

The causes are ordinary. Scales leave the factory with slightly different calibration. One may have been dropped or moved. One sits on a slightly sloped bench. Batteries or power supplies differ. One may read in 1 g steps and the other in coarser steps. Load position on the platform matters on some models. None of this is unusual; what matters is whether your process lets the difference reach the candle.

In a small workshop with one person weighing everything, a biased scale changes every batch equally, and the formula is consistently wrong in a way you can correct once. With two employees using two scales, each batch inherits whichever scale was nearest. The formula now varies from batch to batch in a pattern nobody recorded.

Measure the disagreement properly before arguing about it

One comparison with one object tells you almost nothing. You need to know how the difference behaves across the loads you actually weigh.

These readings are illustrations of the method, not typical figures. The pattern is what you are looking for: a difference that grows with load is proportional; a difference that stays the same in grams is an offset. Neither result tells you which scale is right. For that, weigh a certified mass or a verified reference object on both, and let whichever matches it become the reference.

Worked example: one formula, two employees, two loads

A 25 kg batch at 8% of total weight is 23 kg of wax plus 2 kg of oil. The wax is weighed on a platform scale both employees share. The oil is weighed on whichever bench scale is free.

Employee one uses Scale A, which is verified correct, and weighs a true 2,000 g of oil. The batch is 25,000 g at 8.00%. Employee two uses Scale B, which reads about 15 g low at 2 kg. To make B display 2,000 g, they pour in 2,015 g. That batch is 25,015 g at 2,015 ÷ 25,015 = about 8.06%. Reverse the bias and the batch would sit at about 7.94% (1,985 ÷ 24,985).

Be honest about scale. A difference between 8.00% and 8.06% will not be noticed in a burn test. If this were the only effect, you would note it and move on. The reason to act is that the same disagreement behaves very differently on small weighings.

Where a small disagreement becomes a big one

Consider an additive weighed at 1% of a 1 kg development batch, such as Vybar within its stated 0.5–2% range: 10 g. A 2 g offset between scales is a fifth of that dose. A scale reading in coarse steps may not be able to show 10 g reliably at all. The same is true of a single 200 g test candle's 16 g of oil. The disagreement that was trivial at production scale can make development results impossible to reproduce, which undermines the very comparison you rely on when a production batch fails.

Assign each scale one job and make it visible

Once you know how the scales differ, remove the choice from the moment of weighing. The rule is simple: every weighing type has one named scale, and the batch sheet shows which one was used.

Keep the cross-check alive

A single cross-check describes the two scales on one day. Scales drift, batteries weaken and benches get bumped, so the agreement you measured this month is not guaranteed next month. Fold a short version into the daily routine: at the start of each run, weigh the sealed oil container on both bench scales and write both readings on the scale log. It takes under a minute. If the difference moves outside your agreement limit, you find out before the first batch is weighed, not after a customer notices a weaker candle.

Training matters as much as hardware. Show every new team member the cross-check readings and explain why the fragrance scale is reserved for fragrance. People follow a rule more reliably when they have seen the 15 g difference with their own eyes than when they are simply told which scale to use.

What to do about batches already weighed on both scales

If you discover the disagreement after weeks of mixed use, start with the records. Batch sheets that note the scale can be split into two groups and each group's true load recalculated. Where the sheets do not say, ask the team which scale was used and mark those batches as estimated.

For the numbers in the worked example, the load difference is small enough that affected candles would normally be released after the usual burn test on retained samples. If the cross-check reveals a much larger disagreement, or if the unit mode was ever wrong, hold the batches, burn retained candles from each group side by side at the same cure age, and decide from what you observe, not from the scale readings alone.

Where this page fits

CSI already has guides that cover parts of this question. This page covers the failure investigation for the question above; these go deeper on the rest:

Three things to buy for this job

Frequently asked questions