How to make a glaze

Before you start

Making a glaze means weighing powdered raw materials (ground minerals, clays and frits, which are glass made in a factory and ground to powder) from a recipe, and mixing them with water into a slurry, a thick liquid of powder held in water. Then you sieve it, adjust its thickness, and test it on your own clay before you trust it on a pot. Every number in this guide is a starting point: the right water, thickness and sieve for your glaze are the ones your tests show, so write down what you do and what you get.

What you need

• A scale that reads to 0.01 g for tests and colorants, and a larger one for kilogram batches.
• Containers with lids: low ones for dry materials, buckets for glaze. Derek Au keeps test glazes in lidded soup containers.
• A clean scoop for each material, so one powder never carries into another.
• A mixing tool: a drill with a mixing paddle, or for small tests a blender kept only for glaze. A whisk will do if you have nothing else.
• Sieves, sold in the UK as sieves and lawns: 80 mesh, plus 100 or 120 mesh for glazes with colorants.
• A 100 mL graduated cylinder or syringe, for measuring specific gravity.
• Labels, a waterproof marker and a notebook, or Glazecalc's notes and firing logs.
• Test tiles made from the clay you will glaze.
• A respirator, rubber gloves, a sponge and a mop.

Buying materials

Where to buy

Potters buy from ceramic suppliers, which sell clay, glaze materials and tools. The table shows some for your country, each checked on the shop's own site, and you can choose another country to see its shops. Stock changes, so check before you order. Large producers such as Sibelco and Imerys sell to industry; potters buy their materials through shops like these. The ceramics reference site Digitalfire keeps a directory of supplier stores for other places.

How much to buy

The standard sack is 50 lb (22.7 kg) in the US and 25 kg (55 lb) elsewhere, so the two are nearly the same. Many suppliers also sell 1, 2.5 or 5 kg packs (about 2.2, 5.5 or 11 lb) for tests. Glazy, the open recipe database, advises buying the materials you use steadily (feldspar, silica, kaolin, whiting, and frit for mid-fire) in full sacks, which are often 20 to 30% cheaper. Buy colorants, the metal oxides and stains that color a glaze, in small amounts: most are used at under 5%. Glazy's sample shopping lists came to about $515 for mid-fire and $397 for high-fire, at US prices.

A starter set

Glazy's core list for any temperature is silica, kaolin, feldspar, whiting, zinc oxide, and dolomite or talc. Most mid-fire (cone 5 to 6) glazes also need boron, usually from a frit, which Glazy calls "consistent and less soluble than raw borates". The set below is our suggestion, drawn from Glazy's list and a forum list. The glazing basics guide explains what each material does.

Data sheets and safety data sheets

Ask for two documents for each material. A technical data sheet (TDS) gives its chemical analysis: the share of each oxide in the powder, by weight. Glazecalc works from these, and you can add a material from a supplier's analysis on the Materials page. A safety data sheet (SDS) says how to handle the raw powder. Keep each SDS, and read it before you mix.

Analyses are averages, not promises. Valentine Clays prints "MEAN VALUES. THESE DO NOT REPRESENT A SPECIFICATION" on its flint data sheet, and Glazy's analyses of Custer feldspar from different years run from about 68.5% to 72.4% silica. We suggest writing each sack's lot number (its batch code) in your records, so you can tell whether a change in a trusted glaze came from the material.

HSE warns that crystalline silica "can be wrongly labelled as 'amorphous silica'". And an SDS is about the powder, not the pot: it says nothing about whether a fired glaze is safe for food. Only a laboratory leach test of the fired pot can show that.

Names change, so buy by chemistry

Brand names come and go as mines close. Custer feldspar, mined in South Dakota for more than 75 years, ran out in late 2023. When Clay Art Center switched to G-200 AU and Mahavir, about half its glazes needed adjusting. G-200 and G-200 HP are gone too, and Laguna now imports Mahavir, whose chemistry is close to G-200.

Gerstley Borate, a natural boron mineral in many older recipes, has been scarce since its mine closed in 2000, though New Mexico Clay reported in November 2025 that Laguna was supplying "Gerstley Borate Original". Sources disagree about Gillespie Borate as a replacement. Its maker sells it as pound-for-pound. Berkeley Potters Studio says it "is not a direct substitute", and Digitalfire found it changed a glaze's color. The safe course is to compare the two by their oxides, then test.

Regional names differ too. In North America, flint, quartz and silica mean the same ground silica, and Digitalfire says you probably won't find true flint powder from suppliers. In the UK, flint is its own product: Valentine's is calcined (heated) silica. Both are nearly pure silica, so we expect them to be close, but check the analysis. For kaolin, Glazy says one-for-one swaps, such as EPK for a china clay, "often work just fine, especially when amounts are <10%".

Many recipes name materials that are sold in one country and not another. The table below takes each material in Glazecalc's library that is not sold where you buy, and finds the closest of its kind that is, by their analyses. The figure is how far apart the two are: the grams by which their fired oxides differ, per 100 g of material, added up over every oxide. Within about 15 g, one can usually stand in for the other in a first test. The names carry the makers' and shops' own codes, so you can ask for them by name.

Glazecalc compares materials by what they bring to the fired glaze, not by name. A recipe with a material no longer made says what is used now. Suggest amounts and compare works the amounts out again for a substitute, and Match with what I have makes the recipe again from the materials on your shelf. Both match the fired oxides only, so test a small batch first.

Storing materials and glazes

Dry materials

Keep powders dry, in lidded containers that stay closed when you are not scooping; Glazy advises airtight ones. Choose low containers over deep bins: HSE notes that a deep bin makes you lean in, and dust has further to fall. Keep them on a shelf, so you can mop underneath. Soda ash and Cornish stone take up moisture from the air. One forum potter dries them on a baking sheet at about 50 °C (122 °F), and warns that thin paint buckets can crack in a freezing garage. We suggest keeping soluble materials (soda ash, borax, lithium carbonate, and raw borates such as Gerstley Borate) airtight and indoors.

Mixed glazes

A glaze in a bucket keeps changing. Slightly soluble materials (nepheline syenite, lithium carbonate, Gerstley Borate, colemanite, strontium carbonate, and frits in general) change how it flows over time and can grow small crystals, more so in a hot room. Gerstley Borate gels a slurry, "even in smaller percentages". Brownish water on a settled glaze is dissolved iron: pour it off and replace it with clean water. Also expect:

• Hard-panning: a glaze mixed thick enough to need no additives often settles into a hard layer. Thinning and gelling it helps (see Specific gravity).
• Smells, from microbes feeding on gums such as CMC. A smelly glaze can still be used if it applies well.
• Freezing: keep buckets from it. Thaw a frozen one at room temperature, and check it for cracks.

Don't pour used glaze back into the stock bucket, or dip brushes into it. Before using a stored glaze, stir it, measure its specific gravity again, and sieve it if it has lumps or crystals.

Labels

Label every bucket with a waterproof marker. Zakin, writing for Ceramic Arts Network, gives the glaze's name, its full recipe including colorants, the cone, the date and, in a shared studio, your initials. We suggest adding the target specific gravity and any Epsom salts or CMC. UT Austin color-codes too: orange buckets for low-fire, white for mid-fire. Treat a glaze with no label as one that might contain lead, and don't use it on pots for food.

Weighing

Percentages and batches

A recipe gives each material as a percentage of the dry weight. The main materials usually add up to 100, with colorants often added on top. To make a batch, the amount you mix at once, multiply every number by the batch size divided by 100. For 5000 g, UT Austin multiplies by 50, so a material at 4% becomes 200 g.

On a recipe page, Scale to a batch does the multiplying: enter 500 g, say, and every amount becomes grams to weigh, or pounds and ounces if you choose them under Settings. Print can give just a batch list, with a running total for weighing into one bucket and a box to tick beside each material.

Choose a scale for your smallest amount

A scale's resolution, the smallest step it shows, limits how small an amount you can weigh well. In a 100 g test, a colorant at 0.5% is 0.5 g. A scale that reads to 0.1 g may be off by 0.05 g, about 10% of it; for 0.15 g, the error is about a third. That is our arithmetic, but it shows why colorants need a finer scale.

Advice differs on how fine a scale you need. Derek Au found a cheap 200 g digital scale consistently inaccurate. He recommends a 200 g scale reading to 0.01 g if you buy only one, and rates a good triple-beam balance as "much more trustworthy than a cheap digital scale". The safe choice, in our view, is 0.01 g for tests (50 to 200 g) and colorants, and a bench scale reading to 0.1 g or 1 g for batches of 1 kg (about 2 lb) and up. Check both now and then with a calibration weight.

Weigh in order, and tick each one off

1. Check that you have enough of every material.
2. Put on your respirator. Check that the scale is clean and level.
3. Put the empty container on the scale and press tare, which sets the reading to zero so you weigh only what you add.
4. Weigh each material in turn with its own clean scoop, close each container before you open the next, and tick each material off as soon as it is in.

Derek Au weighs into one bowl in separate piles, so a mistake can be scooped back out. If you weigh into one bucket without taring between materials, Glazecalc's running total tells you what the scale should read after each one.

Mixing and sieving

How much water

Sources disagree. Derek Au starts thick, at 50 mL of water per 100 g of powder, and adds more later. Potterycrafts gives 65 mL per 100 g for one powder glaze. UT Austin uses equal weights of water and powder, and when Digitalfire put 20 kg of powder into 20 kg of water, it measured a specific gravity of 1.46.

Water is quick to add and slow to take out, so start low. Our estimate: begin with about 70 to 80 mL per 100 g (700 to 800 mL per kilogram, or about 11 to 12 US fl oz per pound). Slake, mix and sieve at that thickness, then add water a little at a time until the specific gravity is right. 1 mL of water weighs 1 g, so you can weigh it. If you overshoot, let the glaze settle, then sponge or pour clear water off the top.

Tools and water

• Propeller mixer. Digitalfire calls a good one "essential". A corded drill with a mixing paddle does the job in a bucket. UT Austin runs it at least 4 to 5 minutes after the glaze looks mixed, and keeps the blade off the bucket, which it can scrape plastic from.
• Blender. It "works extremely well for small batches" (Digitalfire). Keep it for glaze only: it must never go back to the kitchen.
• Whisk or stick. Zakin's choice when there is no mixer.

Water varies too, and Digitalfire notes that it changes how a slurry behaves. Derek Au uses reverse-osmosis water because his tap water is hard. Whatever you use, note it.

Step by step

1. Put on your respirator, and rubber gloves if your hands will go in. Zakin wears a mask from the start of weighing until the glaze is wet.
2. Measure about 70 to 80 mL of water per 100 g of powder into the bucket.
3. Add the weighed powder gently, from a low height. UT Austin adds powder to warm water; Zakin and Au add water to powder. We suggest powder into water, which puffs up less dust, except for a recipe with a lot of bentonite: stir its dry powders together first.
4. Slake it: let the powder soak without stirring. Digitalfire waits about 20 minutes. Powders with a gum or binder need to stand overnight.
5. Mix until there are no dry pockets or lumps.
6. Sieve it twice (see below).
7. Cover it and let it rest overnight, then mix again and check the bottom for a hard layer.
8. Measure the specific gravity, and add water a little at a time to reach your target.
9. Label the bucket, and write down what you did.
10. Clean up wet: mop the floor and wipe every surface, including the outside of the material containers.

Sieving

A sieve is a mesh screen in a frame. Pushing glaze through it with a brush breaks up lumps that mixing leaves and catches particles that would fire as specks. The higher the mesh number, the finer the screen: 80 mesh has openings of about 0.18 mm, and 100 mesh about 0.15 mm.

Sources recommend different meshes. Zakin sieves twice through 50 or 80 mesh. Derek Au uses 120 mesh for most glazes, especially those with cobalt or iron: three passes through 80 mesh still left iron specks in one celadon, and one pass through 120 mesh cured it. He keeps 60 to 80 mesh for ash glazes and others with coarse materials. Potterycrafts, in the UK, suggests 60 to 80 mesh. A safe default:

Specific gravity

Specific gravity (SG) tells you how much powder is in your slurry. It is the weight of a volume of glaze divided by the weight of the same volume of water. Water is 1.00, and a glaze at SG 1.45 weighs 145 g per 100 mL. Mixing to a measured SG is how you get the same thickness, and the same coat, every time. It has no units, so it is the same in grams or ounces.

Measure it by weighing 100 mL

1. Stir the glaze so nothing has settled.
2. Put a 100 mL graduated cylinder (a tall, marked measuring tube) or a 100 mL syringe on the scale, and tare it.
3. Fill it with glaze to exactly 100 mL, reading the mark at eye level.
4. Weigh it and divide the grams by 100: 145 g is SG 1.45.

Digitalfire also marks a cup at the level of 500 g of water, fills it to the mark with glaze, and divides the weight by 500; a bigger, fuller measure is more accurate. A hydrometer, a float read where it sits in the glaze, is quick but can misread thick or gelled glazes. UK hydrometers, such as Potclays', read in degrees Baumé (°Bé). By our arithmetic, SG = 145 ÷ (145 − °Bé), so 46 °Bé is about 1.46. Older UK books give a pint weight, the ounces in an imperial pint of glaze: divide it by 20, so 29 oz is about SG 1.45 (also our arithmetic).

What to aim for

Sources disagree, most of all on brushing and spraying, where the UK supplier Potclays gives much higher figures (its own are "for guidance only"). Treat every number as a starting point. Digitalfire's core advice is to find the SG and gelling that work for each glaze, then repeat them and write them down.

To check a dipping glaze by eye, AMACO dips for 5 seconds and looks for a coat about as thick as a US dime. Digitalfire fires tiles dipped for 5 seconds to check.

Water for a target SG: our estimate

The table below is our own arithmetic, not a source's. It assumes the powder's particles have a density of 2.6 g per cm³, which is an assumption, not a measurement. Clay, soluble materials and gelling all change a real glaze, so use it to start, then measure. As a check, it predicts SG 1.44 for equal weights of powder and water, close to Digitalfire's measured SG 1.46. It also shows that 70 to 80 mL per 100 g starts you above SG 1.50, thicker than you need.

Epsom salts, CMC gum and bentonite

Epsom salts make a glaze gel slightly when it stands still and flow again when it is stirred. This is called flocculating it, and a glaze that behaves this way is thixotropic: it goes on more evenly, drips less and settles less. Digitalfire suggests starting at about 2 g per gallon once the SG is right (roughly 0.5 g per liter, our conversion) and tuning the amount each time you mix, since no recipe should fix the dose. We suggest dissolving it in a little hot water first. In a Digitalfire example, a glaze at SG 1.45 dripped, so water brought it down to SG 1.43 and Epsom salts gelled it back to creamy, and it went on evenly. A rough check: stir, stop and watch. A gelled glaze moves on for about 2 or 3 seconds, then stops and springs back a little.

CMC gum (carboxymethyl cellulose) makes a slurry stickier and slower to dry, and leaves a harder coat that holds to the pot. It matters most for brushing glazes and glazes high in frit. Digitalfire's example blends 3.75 g into 1000 g of slurry, and it has to go in with a blender. A glaze with CMC will not gel with Epsom salts, and since microbes feed on the gum, it can change in the bucket and start to smell.

Bentonite, a very fine clay, helps hold the other particles in suspension. Dipping glazes usually rely on 15 to 25% raw clay for this. For a recipe short of clay, about 1 to 3% bentonite helps; Digitalfire says a glaze with more than 15% clay should not need bentonite at more than 1% of the dry glaze's weight.

Test tiles and line blends

Start small

Derek Au works up in three steps: a 50 or 100 g test, a 1 to 2 kg trial if it works, then a 5 to 10 kg studio batch. By our estimate, a 100 g test at SG 1.45 makes about 135 mL (about 4½ US fl oz) of slurry, enough to dip a few small tiles in a narrow cup.

Test on your own clay

A glaze can look quite different on another clay or in another kiln: Clay Art Center tested each of its reformulated glazes on four clay bodies, at cone 6 and at cone 10. Potclays' beginner's guide to glaze testing suggests:

• 6 to 10 test tiles (small pieces for trying a glaze) of the clay you will actually use, noting whether they are bisqued or bone-dry
• one change at a time
• a single dip and a double dip, to see the glaze thin and thick
• every tile in the same kiln load
• a mark on each tile before firing, with a stamp, carving or underglaze pencil.

We suggest tiles that stand upright, since a flat tile hides how far a glaze runs, with a textured band to show how the glaze breaks over edges, and a bare foot. Set each on a small dish or a scrap of bisque, or on a kiln-washed shelf, to catch drips. Put a witness cone, a small ceramic cone that bends when it has had enough heat, near the tiles to show what they received; the firing guide explains cones. Give each tile a short code, and write it against the recipe in your records. Willers, writing for Ceramic Arts Network, uses numbers and letters, and writes the system itself in the front of the notebook.

Line blends

A line blend is a row of tests that steps evenly from one mix to another: from one glaze to a second, or from a base glaze to the same glaze with more colorant. Derek Au, for example, steps iron oxide through a clear base in 1% increments. A volumetric method associated with Ian Currie saves time, because you mix only the two ends:

1. Mix the two end glazes, A and B, as batches of the same size, and sieve both.
2. Add water to the one with less volume until both have exactly the same wet volume.
3. Stir both, then draw each mix with a syringe into its own numbered cup.
4. Stir each cup and dip a tile.

In our worked example, A is 100 g of a clear base, and B is the same with 4 g of red iron oxide added (a 4% addition). Each milliliter of B holds the same share of base as a milliliter of A, plus its iron, so the cups step from no iron to 4% in 1% steps. Each end gives 100 mL to the blend, which a 100 g batch covers by our estimate.

The method needs careful arithmetic, well-mixed ends of even SG, and an accurate scale. Derek Au also blends by weight instead of volume. A triaxial blend does the same with three ends, laid out on a triangle.

When a step looks promising, Save as a copy on the recipe page keeps it as a recipe of its own and leaves the original as it was.

Keeping records

Willers puts the goal well: "enough information to be able to repeat, or avoid, the results you discovered". Write down what you plan to test before you start, and the results when you unload. Potclays adds the application, thickness, clay body and full firing, including ramps, holds and cooling. A ramp is a rate of heating or cooling, such as 60 °C (108 °F) an hour, and a hold keeps the kiln at one temperature for a set time. Note lot numbers, the water you used and any substitution: Clay Art Center's feldspar swap meant adjusting about half its glazes.

In Glazecalc, a saved recipe holds the recipe itself. A firing log records each firing: the kiln, the date, times, temperatures, cones and notes. Notes hold everything else, such as the batch record above and what each tile showed.

Sources

• Glazy: Raw materials
• Glazy: Specific gravity
• Digitalfire: Specific gravity
• Digitalfire: Epsom salts
• Digitalfire: CMC gum
• Digitalfire: Dipping glaze
• Digitalfire: Bentonite
• Digitalfire: Water solubility
• Digitalfire: Gerstley Borate
• Digitalfire: Mahavir potash feldspar
• Digitalfire: Flint
• Clay Art Center: The end of an era, the extinction of Custer feldspar
• Berkeley Potters Studio: What the flux, November 2023
• New Mexico Clay: Gerstley Borate Original
• Ceramic Arts Network: Glaze chemistry 101, a quick course on mixing ceramic glazes (Zakin)
• Ceramic Arts Network: How to keep good records when testing ceramic glazes (Willers)
• Ceramic Arts Network: 8 ways to apply glaze (Hopper)
• Derek Au: Mixing test glazes
• Derek Au: Digital scales for weighing glazes
• Derek Au: Triaxial testing
• UT Austin ceramics wiki: How to mix glazes
• Potclays: The beginner's guide to glaze testing
• Potclays: Glaze hydrometer
• Potterycrafts: Nordic Blue Mist powder glaze
• HSE: Glaze and colour preparation (CR1)
• Valentine Clays: Flint technical data sheet
• Ceramic Arts Daily forum: Ian Currie test tiles