The First Candles From My 20 Kg Batch Have Weak Cold Throw but the Last Ones Are Strong — Could Fragrance Have Settled or Remained Poorly Mixed?

Settled after mixing, or never mixed at all

Your 20 kg batch started with weak cold throw in the first jars and finished with strong cold throw in the last. The question in the title offers two explanations, and they are genuinely different problems with different fixes.

Poorly mixed means the wax was never uniform. The oil went in, some of it blended, and some of it stayed as a richer zone, usually low in the vessel or near the side where it was poured. The gradient existed the moment you stopped stirring.

Settled means the wax was uniform when you stopped stirring and then separated while it waited. As the wax sits and slowly cools, anything that was only loosely held in the blend can drift. Over a long pouring session the lower part of the vessel ends up richer than the upper part. The gradient grew during the hold.

The difference matters for your next purchase decision. A poorly mixed batch needs better stirring. A batch that settles needs a shorter hold, a re-stir before each fill, or smaller scented portions. Buying a mixer fixes the first and does very little for the second if the wax still sits untouched for an hour afterwards.

The 20 kg batch in numbers

At 8% of total candle weight, a 20 kg batch is 18.4 kg of wax and 1.6 kg of oil, which fills 100 candles of 200 g before losses. Suppose the wax was Premium Coconut Soy Wax CSI 468 and the oil was Roasted Coffee.

CSI 468 in the 10 kg pack is ₹6,370.00, which is ₹637.00 per kg, so 18.4 kg costs 18.4 × ₹637.00 = ₹11,720.80. For the oil, a 1 kg bottle at ₹3,738.00, a 500 g bottle at ₹1,890.00 and a 100 g bottle at ₹390.00 give exactly 1.6 kg for ₹6,018.00. The scented wax cost ₹11,720.80 + ₹6,018.00 = ₹17,738.80, which is ₹177.39 per candle across 100 jars (₹17,738.80 ÷ 100).

Now think about time. With the 600 ml Steel Pouring Pitcher, at roughly 0.9 g per ml for melted soy wax, one fill carries a little over 500 g, so a 20 kg batch needs close to forty dips. If each dip, pour and reset takes a minute or two, the last wax in the vessel waited far longer than the first. That wait is where settling happens.

A timed top-and-bottom test that separates the two causes

You cannot go back and sample the batch that already set, but you can run the next one in a way that tells you which failure you have. The test uses small sample pours taken from two depths at two moments.

The sample counts and timings here are suggested starting points. Validate them on your own wax, oil and vessel, because a coconut-soy blend, a container soy and a pearl wax may each behave a little differently during a long hold.

Clues the existing batch can still give you

Even without samples from the vessel, the set candles carry evidence. Weigh one jar from each quarter; if the last quarter is heavier, part of the stronger smell is simply more scented wax per jar. Look at how you filled the pitcher: dipping from the top means the last quarter came from the floor of the vessel, which fits both explanations but rules out neither. Check whether the stirring stopped completely once pouring began. A batch that was stirred between every fill and still shows a gradient points more towards poor initial mixing, while a batch that was left alone for most of the pour points more towards settling.

Roasted Coffee's product page says its vanilla and caramel base may cause slight yellowing in white wax and calls that cosmetic. Over the coming weeks, if the last quarter warms in tone faster than the first, treat it as a weak hint that those jars carry more oil. It is a hint, not a measurement, and it can never replace the blind comparison.

What to do with the 100 candles already poured

Hold the batch. Group the jars by pour order into rough quarters of 25 and rank one coded jar from each quarter against your approved reference candle. The quarter that matches the reference is your release standard; the others are the question.

A weak first quarter that burns cleanly is not dangerous, only disappointing. Sell it only if you describe it honestly or use it for samples and display stock. A strong last quarter needs a burn check before release, because a jar carrying noticeably more oil than intended can burn differently. If any jar smokes, flares or throws a tall flame, stop the test, never leave it unattended, and do not release that quarter.

Keep one jar from each quarter as a retained sample. When the corrected process runs, you will want these for a side-by-side comparison.

Controls that stop a first-weak, last-strong batch

Three habits prevent most repeats. First, add the oil at a temperature below its published flashpoint and never with the melter at maximum; Roasted Coffee's page gives a flashpoint of approximately 191–193°C, and the wax maker's stated temperature governs where you actually add it. Confirm the reading with a separate thermometer, because the CSI melter's display has a stated ±10°C deviation.

Second, stir the whole mass, floor and walls included, and then re-stir briefly before every pitcher fill. Third, cap how long scented wax may wait before it is poured. As a suggested starting point, set a hold limit you can actually meet for a 20 kg batch, record it, and shorten it if the timed test shows settling.

If your pouring pace cannot keep up with 20 kg, scent in smaller portions, say two lots of 10 kg, so no wax waits too long. That costs a little more time at the melter and saves a batch of disappointing first jars.

Write the controls onto the batch sheet as fields to fill in, not as advice to remember: oil-in time, mixing-stop time, re-stir before each fill ticked, last-candle time. A sheet that asks the question every batch gets answered every batch. When a helper runs the melter on a busy day, the sheet is what keeps the method the same as the one you proved.

Only after these controls are in place and the timed test still shows a gradient should you look at mechanical mixing.

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