My Small-Batch Candle Formula Uses 8% Fragrance Successfully but 8% in a 25 Kg Batch Gives Inconsistent Scent Between Candles — Why?

8% for the batch is not 8% in every jar

At 1 kg, 8% is simple: 920 g of wax, 80 g of oil, one pitcher, five jars, and every jar comes from the same well-stirred pot. At 25 kg the percentage is still right on the sheet — 23 kg of Luxury Soy Wax CSI 464 and 2 kg of Jasmine — but the jars now come from a process with more places for the oil to be shared out unequally. Inconsistent scent between candles is a distribution symptom, not a load symptom. The total can be exactly right while individual jars sit well above and below it.

That distinction decides everything that follows. Raising the load will not narrow the spread; it just moves every jar up, the strong ones included. You will pay for more oil and still get uneven candles. The fix lies in finding where in the process the oil stopped being shared evenly.

The most common culprit: one oil weighing, several melts

Many makers producing 25 kg do not have one vessel that holds it all. With a melter sold as 10 kg capacity whose page states it melts 9.5 litres at a time — roughly 8.5 kg of melted soy at about 0.9 g per ml, so weigh one real fill of your own — 25 kg becomes three melts. Each melt must get its own correctly calculated oil. That is where a flawless batch total turns into three different candles.

A worked example

Suppose the wax went into three melts weighed as 7.8 kg, 7.6 kg and 7.6 kg — 23 kg in total. The 2 kg of oil was weighed once, then poured by eye, about a third into each. Afterwards the bottle weights show the melts actually received 600 g, 650 g and 750 g.

The batch sheet says 8%, and it is right. But the first 42 jars are at about 7.1% and the last 41 at about 9.0%. Jasmine's product page gives a soy load of up to 10%, so none of these jars is necessarily unsafe, yet customers who buy two jars will notice the difference. The correct method was to calculate the oil for each melt from its own wax weight — wax × 8 ÷ 92 — and weigh it separately.

Why splitting by eye is so easy to get wrong

Two kilograms of oil in a single bottle looks like a lot, and pouring roughly a third of it three times feels controlled. But the eye judges level, not weight, and the bottle narrows as it empties, so the last pour always looks smaller than it is. A 100 g misjudgement per melt is barely visible and is already more than a full percentage point in a melt of this size. The scale on your bench removes that judgement entirely: tare an empty beaker, pour, stop at the calculated figure, and write the number down.

The same trap appears when the three melts are not the same size. If one melt carries more wax because the melter was topped up, an equal third of the oil is already the wrong amount for it. Per-melt calculation handles unequal melts automatically, which is one more reason to calculate from each melt's own weighed wax.

Other sources of jar-to-jar spread

If your batch really was one vessel with one weighed addition, the spread comes from elsewhere. Work through these in order of how often they catch makers at 25 kg.

Sampling plan: find the spread, then find its source

Before any fix, map the inconsistency. The plan below crosses melt or vessel zone with pour position; three jars per cell is a suggested starting point to validate. If you did not label jars by melt, reconstruct groups from pour order and times on the sheet, and start labelling from the next run — a coloured dot on the base per melt costs nothing.

What to do with an uneven batch

If jars can be traced to melts, you can disposition each group on its own rather than condemning 124 jars together. In the worked example, group 3 at about 9% sits inside Jasmine's stated soy range; burn-test it for soot and flame height and release it if it performs cleanly — just not as the same product as your 8% line, if customers compare. Group 1 at about 7.1% is lean. To bring it to 8% you would remelt its 42 jars, weigh the recovered wax and oil together, and top up by calculation: on the full 7.8 kg of wax the target was 678 g, so the missing oil was about 78 g. After remelting, weigh what you actually recover and scale the top-up to it.

If jars cannot be traced, the choice is harder. Sample more widely, and if the spread is large and unpredictable, the honest options are to rework the whole batch — remelt, combine, mix thoroughly and repour with fresh wicks — or to downgrade. The materials here are worth ₹19,516.20 before jars: 23 kg of CSI 464 at ₹507.40 per kg = ₹11,670.20, plus two 1 kg bottles of Jasmine at ₹3,923.00 each = ₹7,846.00.

The control that keeps 8% in every jar

Calculate oil per vessel, not per batch. Weigh each melt's wax, compute its oil at 8% of total, weigh that oil separately into its own beaker or jug, and record all three melts on the sheet. Label jars by melt. Log first, middle and last pour temperatures. And pour the whole batch on one day, or record the pour date on every group so cure age can be matched.

Weighing per melt adds perhaps a minute to each melt. Set three beakers out before the first melt starts, label them with the melt number and calculated oil weight, and fill each on the scale in turn. When the batch is done, three empty labelled beakers tell you at a glance that every melt got its oil — a simple check that would have caught the worked example above before the first jar was poured.

Only once the spread is inside what you and your customers accept should the process be called validated at 25 kg. If your per-melt figures do not reconcile and you would like a check, CSI answers on WhatsApp at +91 7397976926.

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