Wednesday, 24 May 2017

Sticking Kiln Wash

Sometimes people experience kiln wash sticking to the bottom of their glass. 

You need some understanding of what kiln wash is to know why the wash sticks. It is largely due to the chemical changes in the kaolin at fusing temperatures.

Opalescent glass does tend to pick up kiln wash more easily than transparent, and does it more at higher temperatures. It is the case that at higher temperatures and longer soaks, the kiln wash is more likely to stick to any of the glasses than at lower temperatures and with shorter soaks. This re-enforces the mantra of "low and slow" to avoid problems in kiln forming.

To achieve the same effects at lower temperatures as at higher temperatures, your rate of advance needs to be slower from the slump point to the top temperature.  This additional heat work will achieve the desired effect with a lower temperature.

One kiln wash, Primo, does not contain china clay.  If you use this and it is sticking to the bottom of the glass, you may be firing too high. Try a lower temperature with a longer soak to reduce the kiln wash pickup. 

Wednesday, 17 May 2017

Compatibility Tests

These procedures are based on the observation that glasses compatible with the base glass are compatible with each other. This means that you can test opaque colours’ compatibilities with each other by testing each of them on clear strips.

Annealing test

These tests must be combined with an annealing test.  This conists of putting two pieces from the same sheet of glass together - so you know they are compatible - and firing them along with your compatibility test.

Viewing the results of your annealing through the polarised filters shows whether there is stress left in your annealing.  If there is, the compatibility tests are inconlusive as there is no difference in appearance of stress whether from incompatibility or from inadequate annealing.  Once you have the annealing right, you can then interpret the compatibility tests done at the same time.

Strip test

Cut a strip of base glass ca 25mm wide and as long as convenient for you or your kiln.

Cut clear glass squares of 25mm to separate the colours.

Cut 25mm squares of the colours to be tested

Start with a clear square at one end of the clear strip and alternate colours and clear along the strip finishing with a clear square.

Add a stack of two layers of clear to the kiln before firing. This is to test for adequate annealing. If the annealing is inadequate, then the whole test is invalid.

Test the result with polarising filters. Start with the clear annealing test square. If no stress is apparent, go to the test strip. But if stress is apparent in the annealing test, look to your annealing schedule as something needs to change. Usually the requirement is a combination of a longer soak at the annealing temperature and a slower annealing cool.

To test for compatibility, look carefully for little bits of light in the clear glass surrounding the colour. These are indications of stress – the more light or the bigger the halo, the greater the stress. Really extreme stress appears to be similar to a rainbow, although without the full spectrum.

You can use this test to determine if you annealing is satisfactory for larger pieces. In this case you should use at least 100mm squares. Stack them to the height of your planned project and dam them with fibre board or other refractory materials to prevent spread. Fire to full fuse and anneal. When cool check for stresses.

The tile method looks at compressive factors too.

Cut a 100mm clear tile

Cut two strips of glass 25mm wide and 100mm long for each test

Cut two rectangles of 25 by 50mm of the same glass for the two remaining sides

Cut a square of 50mm for the centre. The glass in the middle is normally the test glass. To be very certain of what has happened you can do the reverse lay up at the same time. You put coloured glass around the outside, but in this case the inside needs to be clear or transparent. At least one element needs to be transparent enough to view the stress patterns, if any. So you could have a clear middle and black exterior, and vice versa.

This test is a more time consuming process and you may wish to use it only for larger projects.

Also look at the use of polarising filters

Charges for Repairs

Repairs always cost more than the owner or artist expects on initial inspection.  The cost is very similar to, or more expensive than, the cost of a new panel if the whole has to be taken apart and renewed.

If it is a repair to part of the window or object, you need to be careful that you do not under price.  The cost elements you need to consider are these at minimum:

  • Glass
  • Materials
  • Time
  • Overheads
  • Travel
  • Installation
  • Contingencies
  • Profit

Glass - and the cost of obtaining it.  Can you obtain the same or very similar glass to the original?  If you can’t, is the client willing to have the repair in different glass?  If you get approval, you need to cost it – whether you already have it or not.  If you do not have it in your stocks, you need to add in the cost of getting it whether that is travel or postal order.  You need to include the time either or both methods involve in the costs.

Materials – The materials you will use in addition to the glass need to be considered.  These include solder, Foil or lead, flux, patina, cleaning materials, etc.

Time - labour and admin. You need to assess how much time it will take to do the repairs.  Then multiply that by your labour rate. You do have one, don’t you?  If not, get down to it and create one. Use steps one and two of this description.   You also need to take into consideration the time to recreate a pattern for the broken area if extensive.

Overheads – If your overheads are not included in your hourly rate, this is the time to include them in the pricing.

Travel = Your mileage rate + time to get there and back.  If you don’t have a mileage rate, look at what your local authority allows.  This will be lower than what businesses allow, but are reasonable, and publicly available.  (At the time of writing the allowance in Scotland is approximately £0.50 per mile.)  It takes time to get to the location, so this needs to be included in the cost too. Of course, if they are willing to bring the item, it reduces the cost to the client.

Installation – If you are expected to install the piece, you need to include travel (there and back at least twice) and time.  You also need to include the estimated time to remove and install a substitute (and its cost) as well as installation of the repaired piece.

Contingencies - All repairs have uncertainties.  You do not always know what the progress of repairing will reveal.  You can agree with the client that any work required in addition to the initial agreement will be notified for the client to decide whether to proceed or not.  However, you can take on the risk. This is what the contingency is for. You need to build allowance for these unforeseen developments.  A 10% to 20% of the total costs addition to the price is sensible if you are taking the risk.

These seven elements added together give you the cost of doing the repairs.  That is the bottom line.  But there is one more element to consider:

Profit – You do expect to get a profit from all this work, don’t you?  If not, why do the repair at all?  You are not a charity.  Of course, you can decide to give away your profit.  Before you do, think about what you have to pay for repairs – to your car, your plumbing, etc.  You deserve some profit on everything you have invested in this craft that you love.  The love will die without profit.

The profit level will depend on your objectives, but will range from 20% (very low) to 100% (what shops charge). If you put your work in a gallery or shop on a sale or return basis, you expect to have to pay at least 30% on the sale.  That should be the minimum basis of your profit level on any repair.

This may all sound like it is too much trouble for a simple repair.  Yes, it does take a bit of consideration to start with.  But once you have established the basic labour, travel, overhead and profit levels, the rest is pretty straight forward.  You will have an idea of how long it takes to do the work, to travel, the glass costs, etc., and the profit level. You only need to multiply by the rates you have established to give you the price.  I should warn you - it will be much higher than you initially thought.

Wednesday, 10 May 2017

CoE and Temperatures

CoE as a Determinant of Temperature Characteristics

What CoE Really Tells Us

The wide spread and erroneous use of CoE to indicate compatibility (it does not) seems to have led to the belief that CoE tells us about other things relating to the characteristics of fusing glasses.  It is important to know what CoE means.  

First it is an average of linear expansion for each °C change between 0°C and 300°C.  This is fine for metals with regular behaviour, but not for glasseous materials where we are more interested in the 400°C to 600°C range.  Measurements there have shown very different results than at the lower temperatures at which CoLE (coefficient of linear expansion) are measured.  In kiln forming we are also interested in volume changes and CoE tells us nothing about that.

Unfortunately, CoE does not tell you fusing or annealing temperatures. 
And not even relative temperatures.  

Some examples: 
  • Uroboros FX90 has an annealing point of 525C compared to Bullseye (516/482C), and to the Wissmach 90 anneal of 510C. 
  • Wissmach 90 has a fuse temperature of 777C compared to Bullseye's 804C.  
  • Another example is Kokomo with an average CoE of 93 which has an annealing range of 507-477C and slumps around 565C. 
  • There is a float glass of a CoE of 90 that anneals at 540C and fuses at 835C.  
  • Artista (which is no longer made, except in clear) had a Coe of 94 with an annealing point of 535C and fuse of 835C, almost the same as float with a Coe of 83. 

These examples show that CoE can not tell you the temperature characteristics of the glass. These are determined by a number of factors of which viscosity is the most important. More information can be gained from this post on the characteristics of some glasses, or from testing and observation as noted in this post .

CoE does not tell you much about compatibility either, since viscosity is more important in determining compatibility.  CoE needs to be adjusted and varied in the glass making process to balance the viscosity of the glass.  Viscosity is described here .

This post and its links describes why Coe is not a synonym for compatibility. 

What CoE REALLY tells us is that we look for simple answers, even when the conditions are complex.  

Wednesday, 3 May 2017

Channels in Jewellery Items

The principle in forming channels in fused glass is to keep the space open with something that will survive the firing and can be easily removed.

You can use kiln washed wire, mandrels, or tooth picks which you can pull out after cooling. These tend to leave a residue of the kiln wash behind. So this is best used on opaque items.

You can use rolled or cut fibre paper, which can be washed out after cooling, leaving a clean hole. This is works well on transparent items.

Both these methods tend to leave bumps over the channel. So you can make a three layer piece. Cut the middle layer short enough to allow the element to keep the hole open (toothpick, cut piece of fibre paper, wire etc.) to be placed with enough overlap of the top layer to catch the bottom layer. In this kind of setup you need to make the top layer a bit longer than the bottom layer. Make sure you are generous in the length of the "hole keeper" so if the glass (now possibly 9mm) does expand you do not trap the material inside.

Of course on a three layer set up like this you could use thin glass which would give you about 6mm of thickness thus eliminating the spread due to volume. In this case you would need to use fibre paper or wire that is about 1.5mm high/thick. It is probably best to have a thin piece of glass on each side of the “hole keeper” to ensure the glass does not retreat due to lack of volume.

You can experiment with a layer of standard and two of thin in various combinations to find the one you like best.

Removing Bubbles

Inclusions and Bubbles

The inclusion of material between two or more sheets of glass has the risk of creating bubbles.  The size of these often relate to the size of the inclusion.  The inclusion can be glass (powders, frits, cut pieces), mica, metals, foils, etc.

The important element in eliminating bubbles is to have a long slow bubble squeeze from the bottom of the forming temperature to the top slumping temperature.  If this is combined with supports at the edges or a fine film of clear powder, it will help reduce the interior bubbles to a minimum.  The supports at the edges may be as small as fine frit (and some use powder over the whole surface).

But, once you have bubbles in the piece, what can you do?

You can drill a hole in the bubbles, or break the bubbles and fuse again, but there will be distortions visible in the resulting piece.

Another method to reduce the effect of bubbles, is to flip the piece and fire upside down to drive the bubbles to the bottom of the piece.  Be careful to use low slumping temperatures to avoid enlarging the bubble.  At the finish, the bubble will still be in the glass but will not be protruding above the top surface.

It may also be possible to combine the two processes.  Drill a small hole in the bubbles and fire upside down.  If you do this you need to place the glass on porous fibre paper, not just Thinfire or Papyrus, to allow the air to be compressed out of the bubbles.  You also need to allow a significant amount of time around the slumping temperature for this to happen.

Once you have fired upside down, you will need to fire polish the surface again. Do not despair at multiple firings.  A lot of people fire their pieces many times to achieve the effects desired.

Wednesday, 26 April 2017

Borax solutions

A borax solution can act as a devitrification spray. That is its usual application in kiln forming.  But it can be used in other ways too.

Borax is a flux helping to reduce the firing temperature of glass. So, it can be used as a medium for powdered mica which can be painted or sprayed onto the glass. It also helps reduce the oxidisation of included metals.

Obtain borax that has no additives. Put a couple of teaspoons into water and bring to a simmer. Remove from the heat and cool. Decant the almost clear liquid off the sediment and you have a saturated solution of borax ready to use. 

If you are really parsimonious, you can add water to the crystals remaining in the pot and heat to get another saturated solution. You could do this until there was no residue, but that would get tedious.

Add a couple of drops of washing up liquid to the solution. This is enough to break the solution's surface tension. It helps to give an even distribution of the solution across the clean glass by reducing the beading of the liquid that otherwise occurs.

You can paint the solution onto the material - glass or metal - with a soft brush such as a hake brush, or you can spray it on with a pump spray container.  Be careful to clean the spray container immediately, as borax crystals form quickly.

Make Your Own Stopping Knife

“Stopping knife” is a traditional term for an oyster knife with a weighted end.  This makes it a multi-purpose tool that manipulate glass, dress lead came, act as a fid, act as a putty knife, and become a hammer.  It also stands up on its own.  I find it the single most useful too in leaded glass panel construction.

This note is how to get from here:

To here:

The process relies on the low melting temperature of lead.  This means that you can use stiff paper wrapped around the handle of the knife to contain the molten lead until it cools.

First you set the oyster knife into a vice and cut two dovetail joints at right angles to each other into the end of the wood handle.  This will insure the lead is firmly grasped by the wood and will not come loose during use.

I do this with a fine bladed saw such as a hacksaw, coping saw or even a dovetail saw.  There are Japanese saws that work very well too, but are not so widely available.

The top of the dovetail joint should be just a millimetre or two off centre. 

The angle should be about 30 degrees from vertical.  Saw down far enough to get a 6mm chisel into the space between the two angled cuts.

Chisel out the wood between the cuts.

Repeat for the second dovetail at right angles to the first.

Now you are ready to prepare the oyster knife to become the stopping knife.

Use paper of more than 90 grams per square metre, such as cartridge paper to form the narrow cone.  Set the knife at a slight angle on the paper. 

Secure the beginning edge to the knife handle with a bit of masking tape.  Mark the paper 5 mm – 10 mm above the top of the handle.  This will be the fill indicator when pouring the lead.  If you over-fill the cone, the stopping knife will be heavy and uncomfortable to use.

Roll the paper around the handle to form the cone.  This cone should be as close to vertical as possible.  A wide based cone will, of course, provide stability, but it will add so much weight as to be uncomfortable to use.  It will also be so wide as be uncomfortable for the palm of your hand.

You can unwrap the paper and start over if the cone becomes too wide.  The key is to start the wrapping just before the handle begins to taper toward the end of the handle.  The other way of looking at it is to attach the paper just as the expanding taper stops.

Try to keep the paper cone as smooth as possible.  This will form the shape of the lead end of the handle.  You want it to be as circular as possible without dents or angles.

Now you are ready for the casting.

I use a small old cast iron pot to melt the lead.  I place this over a camping gas burner to provide the heat.  I promise that I did straighten the stabilising legs before lighting the camping burner.

Put some old lead came into the pot to be melted.  While this is coming up to heat, place your wrapped oyster knife in a vice with heat resisting materials around the site to catch any spills.

Put sufficient lead into the pot, as there will be impurities floating on top and the lead will cool quickly when taken off the heat.  The photo below shows the amount of lead used.  This 100mm diameter pot has lead barely covering the bottom.  You do need enough lead to complete the pour at one go, as a second pouring will not stick to the first adequately.

The photo shows the last piece of came just about to be melted.  This is the time to begin the pour.  If the lead is too hot, it burns the wood creating gases and multiple bubbles splashing hot lead and leaving an unpleasant surface for the tool.  As the last piece of the came melts and leaves its impression as the piece on the left, it is time to pour.

Pour at a steady rate into the paper cone until you reach the height indicator you previously marked in the paper.  When you stop pouring, set the pot on a heat proof surface.  You will notice some smoke and browning of the paper.  That is normal.  This picture shows the effect of the hot lead on the paper once the smoking has finished.

This photo shows the inside of the cone while cooling.  The cooling process will take about an hour.  You will be able to check, by touching the paper, how hot the whole is. 

When the whole is cool, you can unwrap the paper from the handle.

This shows the roughness of the handle end.  This is due to the bubbling from the scorching of the wood and paper.

When the paper is removed and the lead is fully at room temperature you can use a rough file to remove the bubbling and to round the edge of the lead.

The oyster knife has been transformed into a stopping knife and is ready to use.

Wednesday, 19 April 2017

Light and Dark in Designs

Chiaroscuro – This word borrowed from Italian ("light and shade" or "dark") refers to the modelling of volume by boldly contrasting light and shade. 

Glass artists need to be very cognisant of light and dark, both in terms of colour selection and in terms of density. A very thick dense glass of a dark shade of any colour will create a much more intense darkness than glass that is thinner and less dense.

In terms of colour, lighter hues go where the sun shines or where the eye is to be drawn. Pastel shades indicate brightness and light. Within some opalescent and art glasses it is possible to find a shade of colour graduating to white or light yellow. 

Shading can be achieved by using the white areas to indicate where light is falling. A denser dark glass can be used to indicate where light does not fall, or where very little light can filter through. It can also play the part of negative space.

Sometimes, it is useful to use a monochrome scheme to assist in determining where the light and dark should be, as in this pear:

The contrast between light and dark can be used in several ways. Darkness can indicate depth of field or distance when used in a general landscape. Or, it can be used to bring a foreground out, making other elements more vivid.

The key thing to remember in using stained glass is to not be afraid of dark glasses. They can very useful, even if of very odd hues of colour.

Wednesday, 12 April 2017

Firing wire inclusions

Wire and other metal inclusions often cause bubbles to occur around them.  The standard solutions are to add frit to the corners, or powder or fine frit around the inclusions.   You can also flatten the wire or metal to reduce it height. These most often work well.  Sometimes though they don’t eliminate big bubbles around the metals.

In this case think about firing upside down. This is not the whole piece; it is only the inclusion and the bottom layer of glass.  Place the wire or other inclusion on the prepared shelf. It will be most successful if placed on 1mm or thicker fibre paper to allow any trapped air to escape through the fibre.  Place the base glass on top and take to a tack fuse with a bubble squeeze included.  You might even want to consider cutting the base larger than the final piece to be able to cut off the thickened edges and make a more successful piece at the end.

After tack fusing upside down, the inclusion will be imbedded in the glass with an almost flat surface and little in the way of air pockets at the edges.  Clean very well, especially any spalling from the metal and of course, clean the glass thoroughly.  Cap and fuse with a bubble squeeze again.  The bubbles around the inclusion should be minimal if not eliminated.

This method will allow the glass to sink around the glass making a much flatter piece for the capped full fuse. It should also make for a flatter finished piece with many fewer bubbles.

Wednesday, 5 April 2017

Large Bowed Pieces

Occasionally, large pieces in the kiln develop a bow at the end of firing.  The most obvious is when the bow is upwards, but it also occurs that the piece is domed.  This is much more likely to be observed when there are complete sheets, rather than ones interrupted with other design elements which break up the whole sheet.

This is a result of a slight mismatch of compatibility.  One glass is expanding and contracting slightly more than the other.  The bow is always toward the glass which expands the most.  When it contracts, it also contracts more than the other glass, drawing the sheet with lower expansion toward it to form a bow.

This is a form of mild stress.  It can sometimes be seen in large sheets of streaky or flashed glass which are not completely flat.

It is not a fatal flaw.  A piece of this nature can survive many years in that state.  I once had a large window to repaint, because of a football impact.  When re-assembled, it showed that it had been bowed from the outset, almost 90 years before.  It is not in a suitable state for wall pieces or other things that need to be flat, of course.


The remedies most often relate to reducing the stress in the piece.

This of course, relates to the firing schedule.  Increasing the length of the soak at the annealing point is one method.  This combined with reducing the rate of cooling can be effective.

Another method can be employed also.  This is to soak the glass just above the upper strain point of the glass.  This soak should be equal to the one planned for the anneal.  The upper strain point temperature – that point above which no annealing can occur -  is about 40C above the annealing point.  Thus, this soak should occur about 55C above the annealing point of the glass concerned.  Then proceed at a moderate pace to the annealing point.  This rate may be the same as the second stage of the anneal cool (as a starting point). Then anneal as usual for the thickness of the piece.  This method can, of course, be combined with the extended soak and reduced cooling rate as first suggested.

A third method can be employed, if the first two do not work.  This assumes one of the sheets of glass is clear.  Place a sheet of clear on the opposite side of the piece to form a glass sandwich with the two pieces of clear.  Then fire as for a three-layer piece of glass.  The assumption behind this is the same as for toughened glass.  The outer layers will hold the inner layer in compression.  But more importantly, will equalise the slight stress, allowing the piece to remain flat when the firing is completed. This can be used with any transparent glass, but the colour change may not be acceptable.

A fourth method is possible.  Turn the fired piece over and fire, to allow the weight of the glass to overcome the tension of the contraction of the more expansive glass.  This can be successful, but it does retain the stress within the resulting piece.  As such it is not a remedy for the stress, but is a way of flattening.


The place of the glass in the kiln can have an effect too.  If the sheet is near the side of the kiln, there can be a stress inducing effect.  All kilns are a bit cooler at the perimeter than at the interior.  This applies to circular, oval and rectangular kilns.  Rectangular kilns have additional cool spots at the corners.  If the glass is near the capacity of the kiln, the cooler corners can induce this bowing stress to otherwise compatible glass.  The thing to do is to stay about 50mm away from the edges of the kiln when firing large sheets into one piece.


The ideal is to know before firing the large piece whether there will be a problem to overcome. This requires a simple test of the glass to be used.

Assuming the final piece is to be two layers thick of different glass colours, cut a strip of each colour about 50mm wide and as long as the final piece.  Assemble them in the same order as you plan for the final piece.

Add an annealing test square of the two glasses stacked on top of one another.  If one is opalescent and the other is transparent. Make the transparent larger than the other.  If both are opalescent, you will need to run a compatibility test at the same time as this test.  In simple terms, it is to put each of the opalescents on a strip of clear or transparent with the gaps between the opals filled with the transparent.  This test will tell you whether you have fired so fast as to induce stress and so invalidate the test.

Fire as though for a 50mm piece of jewellery – about 200C to bubble squeeze - but without a soak - and then at 400C to top temperature.  Cool to annealing temperature for 15 minutes and cool at 120C per hour to 370C and turn off.

If the long strip is bowed, and the anneal test piece shows no stress, there is enough compatibility mismatch to require the use of one of the remedy methods outlined above for the main piece. It may of course, cause a reconsideration of the glasses to be used or the size of the piece.

Wednesday, 29 March 2017

Colourising Metals for Glass

There are many minerals and chemicals that are used to give glass its variety of colour.  This note attempts to give information on the most common elements and combinations used to impart the colours to the glass.

Antimony oxides produce white glass as do tin oxides.  Together with lead, antimony results in yellow.

Cadmium together with sulphur forms cadmium sulphide and results in deep yellow colour.  Together with selenium and sulphur it yields shades of bright red and orange. 


Chromium is a very powerful colourizing agent, yielding dark green or in higher concentrations even black colour. Together with tin oxide and arsenic it gives emerald green glass. Chromium aventurine, in which aventurescence was achieved by growth of large parallel chromium(III) oxide plates during cooling, was also made from glass with added chromium oxide in amount above its solubility limit in glass.

The material can be introduced into glass either in the form of chromic oxide or potassium dichromate, the latter being a more convenient form.  Potassium chromate is yellow and this colour can be imparted to certain glasses. To produce emerald green glass in which a yellowish cast must be avoided, the addition of tin oxide and arsenic is necessary.

Chromium is associated mainly with the production of green glass, but other colours from yellow through bluish-red, red to dark green or even black can be achieved in combination with other oxides.

Cobalt is the most powerful blue colorant used in glassmaking producing rich blues when used in potash containing mixes, but it can also give shades of green when used with iodides.  The deepest of blues are produced when used in glass containing potash.

Small concentrations of cobalt (0.025 to 0.1%) yield blue glass. The best results are achieved when using glass containing potash. Very small amounts can be used for decolourizing. Addition of 2% to 3% of copper oxide produces a turquoise colour.

Copper is a very powerful and versatile colouring agent when used in colouring glass.  Copper greens and blues are not difficult to produce, although the behaviour of copper in a silicate melt can be complicated.  Copper was used most profusely to produce green glass.  

Pure metallic copper produces a very dark red, opaque glass, which is sometimes used as a substitute for gold in the production of ruby-coloured glass. The art of using copper for ruby glass goes far back to ancient times but even so using copper oxide to make ruby glass can be very difficult. Today we find copper being used to produce turquoise blue tones.

Didymium gives green colour (used in UV filters) or lilac red.

Metallic gold, in very small concentrations (around 0.001%, or 10 ppm), produces a rich ruby coloured glass (gold ruby), while lower concentrations produces a less intense red, often marketed as cranberry. The gold is used as gold chloride. The colour is caused by the size and dispersion of gold particles.

Iron is a very useful and powerful colouring agent even though it can be an undesirable impurity in making glass. Iron in its metallic forms cannot remain in equilibrium with glass and can be disregarded. But its ferrous and ferric forms are of a great help in producing coloured glass.  Iron(II) oxide may be added to glass resulting in bluish-green glass.  

Together with chromium it gives a richer green colour.  In a reduced condition, it can be combined with chromium to produce a deep green glass. 

Used with the combination of carbon or other reducing agents, sulphur and iron sulphides give a dark amber colour.

Lead compounds produce a range of yellows.

Manganese can be added in small amounts to remove the green tint given by iron, or in higher concentrations to give glass an amethyst colour.

Manganese dioxide, which is black, is used to remove the green colour from the glass. This results in a very slow chemical process where it is converted to sodium permanganate, a dark purple compound. Windows made with manganese dioxide solarise to change to a colour which is lightly tinted violet because of this chemical change.

Manganese in its low state of oxidation is colourless, but it is a powerful oxidising agent and can be used for decolourising purposes to oxidise the iron content.  Manganese is mainly used in the production of purple glass resembling the colour of potassium permanganate crystals. The purple colour is achieved by the trivalent manganese however in its divalent state it only imparts a weak yellow or brown colour.

Nickel, depending on the concentration, produces blue, or violet, or even black glass. It is used in the production of smoky coloured glass

Lead crystal with added nickel acquires purplish colour. Nickel together with a small amount of cobalt can be used for decolourizing of lead glass.  When it is introduced into lead crystal it gives a purplish colour, which compensates for a yellow tint produced by other constituents.

Selenium, like manganese, can be used in small concentrations to decolourize glass, or in higher concentrations to impart a reddish colour, caused by selenium nanoparticles dispersed in glass. It is a very important agent to make pink and red glass. When used together with cadmium sulphide, it yields a brilliant red.

Silver compounds such as silver nitrate and silver halides can produce a range of colours from orange-red to yellow. The way the glass is heated and cooled can significantly affect the colours produced by these compounds.

Sulphur, together with carbon and iron salts, is used to form iron polysulphides and produce amber glass ranging from yellowish to almost black.  With calcium it yields a deep yellow colour.  

In borosilicate glasses rich in boron, sulphur imparts a blue colour. Cadmium sulphides, which have a deep yellow colour, are often used in the production of glazes and enamels.

Tin oxide with antimony and arsenic oxides produce an opaque white glass (milk glass), first used in Venice to produce an imitation porcelain.

Adding titanium produces yellowish-brown glass. Titanium, rarely used on its own, is more often employed to intensify and brighten other colourizing additives.

Uranium (0.1% to 2%) can be added to give glass a fluorescent yellow or green colour.  Uranium glass is typically not radioactive enough to be dangerous.  It is often referred to as Vaseline glass by USA collectors.  When used with lead glass with very high proportion of lead, produces a deep red colour.

A good visual chart of minerals and resulting colours is here.

Influence of the glass-making process on colour.
It is not only the minerals that give the glass the colour; it is combined with the way in which the materials are treated.  The physical conditions under which the glass is made also have an influence on the colour.  The main ones are:
1.   The temperature of the melt/batch
2.   Temperature of reheat during the working of the glass
3.   The temperature of the 'Lehr' (Annealing Oven)
4.   Duration of the melt/batch
5.   Time and temperature relationship at different stages in production
6.   The type of colorant being used
7.   Concentration of the colorant
8.   Atmosphere of the furnace
9.   The composition of the colorant within the glass composition, as is the case when iron is added to glass. The type of iron oxide formed decides if the glass will be blue or yellow
10.                The number of times the same glass is melted. Repeated melting of the cullet will usually give a darker tone to the finished piece