Why Did My Glass Crack? A Diagnostic Guide for Lampworkers

Short answer: When your glass cracked tells you why. A crack in the flame is almost always thermal shock — cold glass introduced too fast, or a thick-thin junction heated unevenly. A crack during or right after the kiln cycle points to a bad annealing schedule (or no annealing at all). A crack that appears days or weeks later, often a clean split, is the signature of locked-in stress from incompatible glass or skipped annealing. Each cause has a different fix — and re-annealing only helps an intact piece; it will not heal a crack that has already opened.

Losing a piece stings, but cracked glass is one of the most diagnosable failures in the craft: read the timing, match it to the cause below, and change one thing next session.

Read the crack first: when it happened tells you why

Cracks from missing or bad annealing and from incompatibility can appear after days, weeks, or even months — and community experience holds that unannealed beads more often split cleanly in half than shatter. That clean, late split is a diagnostic clue in itself.

It cracked in the flame: thermal shock, cold glass, and thick-thin junctions

Glass expands when heated and contracts when cooled. Heat one part of a piece much faster than its neighbor and the two regions fight until something gives — that’s thermal shock, the number-one killer of glass in the flame. The classic triggers:

• Cold glass introduced too fast. A room-temperature rod plunged into the hot part of the flame will often crack or send a chip flying. Introduce glass at the tip of the flame and warm it gradually before moving into the working zone.
• Thick-thin junctions. A thin stringer attached to a thick gather, a handle on a heavy sculpture, the neck of a vessel: thin sections heat and cool far faster than thick ones, so the junction is where the stress concentrates and where the crack starts.
• Letting part of the piece go cold. On larger or assembled work, the section outside the flame cools while you focus elsewhere; when you swing the flame back, the sudden reheat shocks it.

Glass type matters here. Borosilicate is far less sensitive to thermal shock than soda-lime (soft) glass — one reason boro is forgiving for sculptural work — but boro still shocks if you abuse it, and soft glass demands genuinely patient warming. (See soft glass vs boro vs quartz for the full comparison.)

Your torch shapes this failure mode too. Small beginner burners like the Nortel Minor, Carlisle Mini CC, or GTT Cricket actually force good habits — their smaller flames make you introduce rods slowly — but they’re underpowered for thick boro, and pushing heavy glass on a small flame is a common thermal-shock setup: you can’t keep the whole mass warm, so part of it is always cooling. For large or assembled work, many artists keep a hand torch such as a Nortel Ranger, GTT Sidewinder, or Bethlehem Sharp Flame hand torch playing over the rest of the piece so junctions never go cold during assembly. Flame character matters as well — see soaking vs penetrating flame.

It cracked during or right after the kiln: annealing gone wrong

Annealing is the controlled slow-cool of a finished piece. As the Corning Museum of Glass puts it, hot glass that cools too quickly is highly strained by the time it reaches room temperature, leaving it vulnerable to fracture from minor mechanical or thermal shock. If the piece cracked in the kiln, on opening it, or within hours of skipping the kiln entirely, the cooling path is your suspect. Manufacturer reference points:

• For clear COE 33 borosilicate, Glass Alchemy gives an annealing temperature of 1050 °F and a strain temperature of 960 °F (corroborated by Northstar’s chart).
• Northstar’s boro schedule doesn’t just soak and switch off — it steps down: soak at anneal-minus-125 °F for 50% of the anneal time on pieces 0.25 in thick or less (100% for thicker pieces), then soaks at anneal-minus-200, -350, and -550 °F of 25% of the anneal time each.
• Soft glass (COE 104) soaks are commonly run around 940–970 °F, but brand matters: CiM notes their Messy Color Clear “generally cracks” when annealed below about 950 °F, and a CiM rep has suggested 990+ °F to resolve incompatibility-like cracking.

The common failure modes: a soak too short or too cool for the piece’s thickness, a ramp-down too fast after the soak (crash-cooling through the strain range undoes the soak), or no kiln at all. Boro’s thermal-shock tolerance tempts people to skip annealing, but Northstar is blunt that kiln annealing is still crucial to the integrity of the work, especially thick sculptural or assembled pieces.

One more subtlety: hotter is not automatically safer. Glass Alchemy warns that holding cadmium (opal/crayon) boro colors above 1050 °F for long periods can burn them out — so the answer to cracking is the right schedule, not simply a hotter one. For full programs by glass type and thickness, see our annealing schedules guide.

It cracked days or weeks later: COE mismatch and locked-in stress

The most demoralizing crack appears in a finished piece sitting quietly on a shelf. Two mechanisms produce it, and both are invisible until they aren’t:

COE incompatibility. Every glass expands and contracts at its own rate (its coefficient of expansion). Fuse two glasses with mismatched behavior and they cool locked together at the seam — one wants to shrink more than the other, so permanent tension builds along the junction. The piece can survive the kiln looking perfect, then let go days, weeks, or months later when a temperature swing or a bump pushes the junction past its limit.

Residual stress from bad or missing annealing. Same mechanism, different origin: the strain frozen in by uncontrolled cooling sits in the piece until a minor thermal or mechanical shock finds it.

Even within a single COE family there are traps: Northstar notes that different colors with matching COE can have different annealing characteristics, and that mismatch alone can make a piece check during cooling or later. Each boro color is its own chemical formula and behaves slightly differently.

COE and compatibility: why matching numbers isn’t the whole story

Beginners are often taught “match the COE number and you’re safe.” It’s a useful first rule, but matching COE numbers is not an accurate measure of compatibility: Bullseye’s TechNote 3 explains that viscosity characteristics matter as much as expansion, which is why Bullseye tests every batch for compatibility under polarized light — fired to 1500 °F with a 15-minute anneal soak, then checked for stress — rather than rating glass by a COE number, and why they don’t test against other manufacturers’ glass at all.

Practically: stay within one manufacturer’s tested-compatible line when you can, treat “same COE number, different brand” as unproven, and when in doubt, test — which brings us to the polariscope.

Seeing stress before it cracks: the polariscope test

You can see locked-in stress with two crossed polarizing filters (a polariscope, or two pieces of polarizing film). Stressed glass rotates the polarization of light passing through it: between crossed filters, unstressed glass stays dark, while stress shows as light — an incompatible inclusion or junction glows with a bright halo ring around it.

The classic chip test for an unknown glass: fuse a small piece of it to a known-tested glass, let it cool fully, and view the sample between crossed polarizers. A halo of light at the junction means stress — don’t combine those glasses in real work. No halo means the pairing is safe to use. It’s a cheap, fast habit that saves finished pieces from the shelf-crack fate.

Prevention, cause by cause

Can this piece be saved?

It depends on whether the crack has actually opened:

• Stressed but intact (it glows in the polariscope but hasn’t cracked): yes — re-annealing relieves internal stress in an intact piece. Run it through a proper full schedule.
• Visibly cracked: re-annealing alone will not heal it. A visible crack must be melted back together in the flame, and then the whole piece re-annealed.
• Flame repair is possible but genuinely risky. The accepted method: soak the piece in a kiln at annealing temperature (about 940 °F for soft glass, 1050 °F for boro), reintroduce it to the flame very gradually, and heat the glass adjacent to the crack so it flows into the gap — never blast the crack directly. Lampworkers report roughly 40% success on beads, and the piece can fly apart in the flame, so wear your eyewear and decide whether the piece is worth the gamble.
• Incompatible glass combinations: not saveable in any lasting way. The stress is built into the material pairing; re-annealing or repair just delays the same failure. Chalk it up to the chip test you’ll run next time.

Key takeaways

• Timing is the diagnosis: in the flame = thermal shock; at the kiln = annealing problem; weeks later = compatibility or locked-in stress.
• Warm glass gradually and keep thick-thin junctions evenly heated — boro is more forgiving than soft glass, but nothing is immune.
• Anneal everything that matters, with a schedule matched to the glass and thickness — and don’t crash-cool after the soak. Hotter isn’t automatically safer for every color.
• Matching COE numbers doesn’t guarantee compatibility — viscosity matters too, which is why manufacturers test with polarized light.
• Two polarizing filters reveal stress before it cracks; the chip test vets any unknown glass.
• Re-annealing saves stressed-but-intact pieces; it cannot heal an open crack. Flame repair is possible but roughly a coin flip on beads, and incompatible combinations aren’t worth repairing.

Sources

• Northstar Glassworks, “Annealing” (boro schedules, color behavior, thermal shock) — https://northstarglass.com/annealing/
• Glass Alchemy, “An Introduction to Modern Lampworking” (COE 33 anneal/strain points, cadmium colors) — https://glassalchemy.com/blogs/the-formula/glass-blowing-an-introduction-to-modern-lampworking
• Bullseye Glass, TechNote 3: Compatibility of Glass — https://www.bullseyeglass.com/technotes-3-compatibility-of-glass/
• Corning Museum of Glass, “Annealing” definition — https://allaboutglass.cmog.org/definition/annealing
• The Crucible, “Lampworking/Flameworking guide” — https://www.thecrucible.org/guides/lampworking-flameworking/
• Creation is Messy, Messy Color Clear notes (soft-glass anneal temperatures) — https://www.creationismessy.com/color.aspx?id=24
• Glass Tips, “Mending a Crack” (flame-repair method) — https://glasstips.blogspot.com/2024/01/mending-crack.html
• Wikipedia, “Annealing (glass)” — https://en.wikipedia.org/wiki/Annealing_(glass)

Editor’s note: annealing temperatures and schedules vary by manufacturer, glass formula, and piece thickness — the figures here come from the makers cited and are reference points, not a program for your kiln. Where sources differ (e.g., soft-glass soaks from ~940 °F to 990+ °F), we’ve given the range. Always follow your glass manufacturer’s published schedule and your kiln maker’s instructions.