Melting & Annealing
Melting &
annealing.
The hands-on method we teach on the phone for melting clean glass — plus the annealing math and official guides, straight from Glasma.
We can't give you one recipe. Too many variables.
Every furnace melts a little differently, and the dials that matter are different in every shop. So instead of a number to copy, we'll give you a method that works in any furnace: charge it, watch it, take samples, and read the glass itself. The furnace tells you when it's done. You just have to be there to listen.
Charge to full. Hold hot until the bubbles stop changing. Then drop the temp and let the glass squeeze them back in.
The rest of this page is just that idea, slowed down — with the test gathers and photos that let you see it happening instead of hoping.
Charge it. Watch it.
Read the glass.
Bring the empty furnace to temp
Turn it up to 2200–2250°F and let it get all the way there before you do anything. On most furnaces that's about an hour. You want the chamber soaked at temperature, not just the thermocouple reading hot.
Add the first charge — in the center
Drop a mound of batch into the furnace. Try to keep it off the walls of the pot or liner. Cold batch sitting against the refractory isn't ideal, and the fluxes in unmelted batch can, in theory, accelerate how fast that refractory wears. So aim for the middle.
Then wait for it to flatten, and resist the urge to rake or disturb it. (Opinions differ here — some melters stir, some never touch it. Starting out, leave it alone.)
However long that first charge takes to go flat is your benchmark for how long this furnace needs to heat through a charge. Write it down.
Build up to full, one charge at a time
Once a charge flattens, add the next one. Repeat until the furnace is full. Keep an eye on that flatten time — it's the rhythm of your furnace.
Remember that batch loses a lot of volume as it melts and the gases escape. To end up with a genuinely full furnace, your last charges will often sit mounded well above the top of the crucible or liner.
Take your first test gather
After the final charge goes in, wait about 90 minutes — less for a small furnace, roughly the same flatten time you've been seeing. Then pull your first sample: plunge a punty rod all the way to the bottom if you can, gather, and pull it out to inspect.
Drop a sample on the marver, flatten it, and photograph it. This is sample one.
Sample and photograph every hour
From there: every hour, take a gather, make a flat sample, take a photo. Be consistent — same depth, same spot on the marver, same lighting. Consistency is what lets you actually compare frame to frame.
Depending on furnace size, temperature, thermal mass, heat-source placement and vessel, expect roughly 3–5 hours at 2250°F before the bubbles — their number and their size — stop changing from hour to hour. When the picture stabilizes, you're ready for the next move.
Now drop the temp — the squeeze
When the bubbles have stabilized, turn the furnace down, roughly to 2000°F. Keep taking your hourly gathers and photos. You should see the bubbles shrinking each hour.
The squeeze isn't more cooking
Dropping the temperature is a chemical step, not a waiting game. At the lower temperature the gases get reabsorbed back into the glass — the glass literally squeezes them out of existence.
Most people try to bake the bubbles out by holding at 2250°F longer. It doesn't work. In our experience it actually makes them harder to remove. Drop the temp, wait an hour, keep sampling, and watch them go.
One melt, photographed
hour by hour
This is exactly the method above, documented by a customer on a single melt. Read it left to right, top to bottom — sample one is top-left, and each frame is an hour later. The dense fog of bubbles at the start thins out steadily, and by the final frame the glass is essentially clear.
"Started to take the samples an hour and a half into the cook cycle, and then every hour. The center picture is when the cooking stopped and the temperature crashed — then the bubbles fade. There's no picture of the glass the next day, but it was perfect."
For context, here's the actual program Boyd landed on for his furnace. Treat it as one real-world data point, not a setting to copy — your furnace will want its own numbers. Notably, he cut out a slow four-hour turn-down entirely and the glass was still perfect.
Feed it your
own glass.
Clean, clear cullet — the glass you recover from your own knock-offs, moiles and broken pieces — is the cheapest raw material in the shop. Reusing it is good economics, and used right it actually helps the melt in two ways.
Cap the mound with it. A small amount of cullet laid over the top of a fresh batch charge can, in theory, help hold the more volatile components in as it heats — for a cleaner, more complete melt.
Use it to tune the colour. Glasma intentionally adds erbium to the batch so your cullet still works even when it carries a little iron. One side effect new melters notice: a fresh melt of pure batch often comes out with a subtle pink-purple tint. Some makers love it — Dylan Martinez leans into exactly that cast in his hyperreal water-bag pieces. If you'd rather it gone, melt in some of your own cullet; the trace pipe scale it carries quietly cancels the tint out.
One step you can't skip: get the dust out
When you recycle cullet from knock-offs, moiles or broken work, make sure it's free of fine dust. If it's crushed, rinse it with water to wash the fines out first. Skip this and those fines will seed bubbles in your melt that are very hard — often impossible — to clear.
You can push it.
But you have to babysit.
Once you know your furnace, you can run hotter to speed things up. The trade is attention: hotter melts move fast and they don't forgive a missed check. Stay on your gathers and your photos.
Don't overcook the glass
Going faster only works if you're watching. Push the temperature without observing and you can cook past the sweet spot — which costs you more than the time you were trying to save.
Bubbles first.
Cords next.
Clean, bubble-free glass comes first — follow the process above and you'll get there. Cords are the subtler problem: streaks of off-composition glass, a matter of mixing rather than heat. Some melters chase them hard, some never give them a thought. When you're ready, the cords guide breaks down what they are, where they come from, and how to chase them out.
Cords
The companion field guide to this one — reading your glass for cords, the studio causes, and the levers that actually move them.
Cool it down right,
or it cracks.
Once a piece is formed it has to cool slowly and evenly, so internal stress doesn't build up and split the glass later. Annealing runs in three stages — stress release, the anneal soak, then the final cool — and each stage's temperature and duration depends on the glass type and how thick the piece is.
We're not going to reinvent that math. Glasma keeps a free calculator that gives you the full schedule for your exact product and thickness. Use it.
Glasma Annealing Calculator
Enter your Glasma product and glass thickness and get a stage-by-stage annealing schedule. Free, and kept current by Glasma.
The official guides
Glasma has supplied glassmakers since 1979 and maintains the authoritative specs, melting recommendations and datasheets. Rather than copy them here where they'd drift out of date, we point you straight to the source.
Stuck on a melt?
This is the same walkthrough we give on the phone. If your furnace is fighting you, send us your numbers and a couple of sample photos — we'll help you read them.
Email us your melt