Cords

Field Guide · Cords

Cords.

Cords are the subtlest thing that can go wrong in a melt — and they’re a mixing problem, not a heat problem. Here’s how to see them, read them, and chase them out.

Scroll — the subtle half of melting
Start here

A cord is just glass that didn’t fully blend with the glass around it.

Inside every melt there are little regions where the composition is slightly off — a bit more silica here, a little less soda there. Because molten glass flows like cold honey, those off regions get stretched into long, thin ribbons. A ribbon like that bends light a little differently than the glass around it, so you see it: a string, a ream, a watery ripple in the finished piece.

It isn’t dirt and it isn’t a bubble. It’s a blending problem — which is good news, because blending is something you can control with heat, time, and how you treat the melt.

Sharp / wire cord
A hard edge you can almost trace with a fingernail. Something foreign drained into the melt — usually your pot or the crown above it.
Smooth / wave cord
A soft optical waviness, no clear line. The melt itself never homogenized — a grain, mixing, or time problem.
Specks on a cord
Tiny crystals or grit riding a streak. Almost always a tell that the cord came from worn refractory, not the batch.

Adapted for the studio from the melting and “Cords and Bottom Glass” chapters of Scholes & Greene, Modern Glass Practice (7th rev. ed.) — the standard industrial reference, pulled out of engineer-speak.

The usual suspects

Where cords come from

Five sources, roughly in the order a studio runs into them. Most of what you’ll ever see is the first three.

A note on vessels: some studios melt in a pot or crucible, some in a tank with a liner. We’ll just say “the pot” throughout — the ideas apply the same way whichever you run. (Which vessel material resists cord best is its own question, and one we’ll tackle once we have solid data.)

1

A worn pot

At melting heat, your vessel — and the crown above it — slowly dissolves into the glass. That dissolved refractory is a different composition, so it shows up as cord. The sharpest, ugliest cords come from material dripping or running in from above the glass line, not from the walls.

You can usually watch it happen: the pot walls visibly degrade over a campaign, and any material missing from the pot is now in your glass. Cords that appear or worsen late in a pot’s life — sharp, sometimes carrying tiny crystals or a faint color — are the pot telling you it’s wearing out.

Sharp, streaky, late in the campaign? Suspect the pot before you blame the batch.
2

Bottom glass

The refractory-rich glass is heavier than the rest, so it sinks and pools in the bottom of the pot. It’s harmless at first — but the layer keeps growing over a campaign, and once it’s deep enough a low gather drags it up, where it shows as a streak low in the gather.

Industrially, bottom cord often carries a green iron tint. With a well-calibrated batch like Glasma, though, the iron is neutralized — so you’re unlikely to see color. It’ll read as optical distortion, not a green streak.

It builds up whether you like it or not — the move is to use it up, not ignore it. See “Manage the bottom glass” below.
3

Incomplete mixing

Early in the melt, the first liquid that forms is runny and tends to run away from the sand grains that haven’t dissolved yet. If the sand is coarse or the batch isn’t evenly mixed, you bake in composition differences that later stretch into soft, wavy cord.

This is the cord you fix at the source. Even, fine, fully-mixed batch never makes it — which is a big part of why pellets melt cleaner than loose batch.

Most wavy cord is decided in the first hour of melting, before you ever gather.
4

Surface scum

At melting heat, fluxes slowly evaporate off the top surface, leaving behind a tougher, higher-silica skin. Fold that skin back into the melt and you get thick, ropy cord. Skimming it off before you work removes the problem.

Good news for you: soda-lime studio batch loses far less to evaporation than lead or borosilicate glass, so this is a smaller headache in a studio than it is in industry.

Skim the skin off the surface before your first real gather.
5

A batch error

The brutal one. A missing or mis-weighed ingredient cords the entire melt, top to bottom, with no fixing it. The defense is boring: trust your batch, don’t freelance the recipe mid-campaign, and change one thing at a time so you always know what moved.

The thing nobody tells you

Your furnace is already stirring

Heat one side of a pot hotter than the other and the hot glass — being lighter — rises, rolls across the top, and sinks down the cooler side. That slow loop runs the entire time you’re melting, quietly folding the glass over on itself. It’s free mixing, and it’s the single biggest reason a steady melt comes out homogeneous without anyone stirring it.

Which means the things that wreck homogeneity are the things that kill that loop: yanking the temperature up and down, never giving the melt time to circulate, working a furnace that’s barely at heat. Steady heat and time aren’t just for chasing bubbles — they’re how cords blend away too.

You can lean on it and add to it. A hotter, longer hold thins the glass and speeds the loop. And here’s a useful quirk: clear glass circulates faster than heavily colored or opaque glass — color and opacity trap heat near the surface and slow the churn, so darker melts often need more time, or a hand-stir, to even out. When convection alone isn’t enough, stirring the melt by hand is the oldest homogenizer there is — more on that below.

See it before you fix it

Read your glass for cords

Cords need a different sample than bubbles. A bubble you can spot in a quick flattened blob — but a cord you have to look through, so you need a clear window of glass with some thickness and clean surfaces. A flat marver sample won’t cut it: chill marks and marver texture scatter light and hide the very distortion you’re hunting for.

Make a consistent test shape instead — a cylinder roughly an inch across, fire-polished smooth on the sides, with no chill marks or surface texture. Then look through it against a plain background and slowly rotate it. Cords show up as lines, or as a watery ripple that slides as you turn the piece. Bubbles stay put; cords swim.

Pay attention to where the cord sits. A streak low in the gather points to bottom glass; tiny specks riding a sharp line point to the pot. Make the same shape the same way every time and shoot it in the same light — that’s what lets you compare one melt to the next instead of trusting your memory.

A punty / sample rod
A 1″ test cylinder
Fire-polished, no chill marks
A plain background
Your phone camera
A simple log
The levers you actually have

How to fight cords

1

Heat and time — the homogenizing soak

Hotter glass is thinner, mixes faster, and drives a stronger convection loop. A good hot, steady soak does for cords roughly what it does for bubbles: it gives the melt the time and temperature to blend itself before you drop to working heat.

If a melt looks streaky, more steady soak beats more fiddling.
2

Stir or rake it — carefully

When heat and convection aren’t enough, mixing the melt by hand is the most direct fix there is — it’s exactly what industry does for its most demanding glass. A few slow passes with a rod fold off-composition glass back into the body. Two cautions: melters disagree on how much to disturb a melt, so start gentle; and your stir tool has to be clean and non-reactive, or a dirty, eroding rod will add cord faster than it removes it.

There’s also a time-honored studio version of what industry calls bubbling: jam a potato on the end of a rod and plunge it into the melt during the melt phase. It throws off a fast burst of large bubbles, and the thinking is those big bubbles both stir the glass and sweep up smaller bubbles as they rise — helping cords and seeds at once. It can genuinely help, but it’s a trial-and-error move: timing and amount depend on your furnace and glass, so test it on a melt you can afford to learn from.

Slow, clean passes — you’re folding, not whipping. (The potato is the exception: there, big bubbles are the whole point.)
3

Manage the bottom glass

The heavy, refractory-rich layer on the bottom never stops growing, so the goal isn’t to avoid it — it’s to use it up and replace it with fresh. Periodically draw the bottom down, working it into pieces where a little cord won’t show, then top back up with new batch. (Industrial tanks literally drill a hole and drain the bottom layer off; a studio just gathers it out on purpose.)

Don’t let it pile up for a whole campaign — cycle it through on forgiving work.
4

Skim the surface

Pull the scummy skin off the top before you work. Thirty seconds with a rod removes a whole category of ropy cord.

5

Start clean

Good batch, right grain, well mixed. Most wavy cord is baked in during the first hour of melting, so even, fine, fully-mixed batch is the cheapest cord insurance there is.

6

Watch the pot — and make it last

Sharp, discolored, speckled cords late in a campaign, plus visibly degrading walls, mean the vessel is wearing through. No melt schedule fixes a spent pot — when the cords say it’s done, change it. To stretch pot life in the first place, heat slower and lower: gentler ramps and a lower peak temperature are easier on the refractory and buy you more melts before it starts shedding into the glass.

7

Be steady — the pro secret

Glass plants live by one rule: don’t make big changes fast. Move temperature gradually, change your batch or routine in small steps, and — this is the counterintuitive part — when cords suddenly appear, the right move is often to hold everything steady and let the melt settle, not to start yanking dials.

Frantic fixes cause more cords than they cure.

When in doubt, change less. Give it a melt or two to clear.
Become the expert

How to experiment

You don’t need a lab. You need a method — and it’s the same one scientists use: change one thing, hold everything else, give it time, and write down what you saw.

1

Change one variable at a time

Hotter soak or longer soak or skim vs. don’t or gather depth — never two at once, or you won’t know which one moved the needle.

2

Give it a full melt

Glass responds slowly. Judge a change over a whole melt cycle, not an hour. A melt is one data point — not proof yet.

3

Sample and photograph the same way

Make the same test cylinder every time — same size, fire-polished, no chill marks — and shoot it through, in the same light and against the same background. A photo log turns “I think it’s better” into something you can actually compare side by side.

4

Keep the winners

When something works, lock it in before you test the next thing. That’s how a method — your method, tuned to your furnace — gets built.

Good first experiments to run, one at a time:

Longer hot soak
Higher peak temp
Skim vs. no skim
Shallower gathers
Fresh vs. worn pot
Finer vs. coarser batch

Some cord is normal

Every melt has a little cord. The real question isn’t “zero cord” — it’s whether it’s below the level that matters for the work you make. Some artists chase cords hard; plenty never think about them. Decide what your work actually needs, then melt to that. Chasing a perfection you don’t need just costs you time and fuel.

Got a cord you can’t crack?

This is the same walkthrough we give on the phone. Send us your numbers and a couple of sample photos — we’ll help you read what your furnace is telling you.

Email us your melt