Your glass looks fine when you walk away from the pour station. Six hours later, a hairline crack runs the full height of three jars in the cooling rack. The pour temperature was in spec. The wick was centered. Nothing changed between “fine” and “broken” except time, and time is where the exotherm finishes its work.
The Exotherm You Cannot See
Paraffin wax releases roughly 220 joules per gram as it transitions from liquid to solid. A 12-ounce jar holds about 320 grams of wax, which means the solidification phase dumps nearly 70,000 joules of heat into a 3-millimeter glass wall over the course of two to eight hours. That is the same thermal load as a small cup of coffee cooling on the table, except the heat is generated inside the glass rather than inside a mug.
Soy wax behaves differently. Its crystallization onset is slower, and the latent heat release is spread over a wider temperature band, which is why pure soy cracks less often than paraffin in the same vessel. But a 70/30 soy-paraffin blend carries enough paraffin to produce the same delayed exotherm. In our factory runs, the highest crack rate hits the 70/30 lines during winter production, when the ambient floor is cooler and the gradient through the glass is steeper.
The 3-Millimeter Gradient
The mechanism is a thermal differential, not a pour error. The outer surface of the glass, in contact with room air at 70 degrees F, sits at maybe 72 degrees F. The inner surface, bathed in the crystallizing wax, is at 100 degrees F or higher. Across a 3-millimeter wall of soda-lime glass with a thermal conductivity of about 1.4 watts per meter-kelvin, that 30-degree delta produces a strain of roughly 0.5 percent.
Annealed glass tolerates about 0.3 to 0.7 percent tensile strain before it yields. The outer surface is in tension. If the glass carries any micro-abrasion from the annealing line, or if the wall drops below 2.5 millimeters at the base where the curvature concentrates the stress, the crack initiates at the weakest point and propagates under the residual strain. This is why the crack appears hours after pour, not at pour time. The peak exotherm in standard 4640 paraffin hits one to three hours after crystallization onset at around 125 degrees F. The glass is still under load when you think it is done.
Pour Lower, Thicker, Warmer
Three levers, in order of impact on the defect rate.
First, lower the pour temperature. A 175-degree pour into a room-temperature jar produces a steeper initial gradient than a 145-degree pour. In our runs, dropping the pour from 175 to 150 degrees F cut the six-hour crack rate from 3.2 percent to under 0.8 percent on a 70/30 blend in 12-ounce glass. The trade-off is slower set time and slightly less wax adhesion at the top edge, both manageable with a 15-minute lid-off rest before capping.
Second, specify a minimum wall thickness of 3.5 millimeters at the base. We switched from a 3.0 mm to a 3.5 mm base on our 12-ounce line in March. The crack rate dropped from 1.4 percent to 0.2 percent on the same paraffin blend. The extra glass costs four cents per unit at our volume, which is cheaper than the return processing and the reputation hit when a customer gets a cracked candle.
Third, pre-warm the empty jar to 85 degrees F before pouring. This cuts the peak differential from 30 degrees to under 15 degrees at the moment of highest exotherm. We run this through a convection oven set to 85 degrees F, two minutes per tray of twenty jars. On the winter batches it is the difference between a 2 percent defect rate and a 0.3 percent rate, and it takes no extra labor beyond loading the oven.