Bubble Squeeze for Multiple Layers

Difficulties often occur with bubble formation within pieces composed of several layers. There are a couple of factors in addition to the number of layers that have an influence - temperature and rate of advance to the bubble squeeze temperature.

Temperature

The top temperature for the bubble squeeze does not need to change with multiple layers. It is the advance to the bubble squeeze that needs to change in relation to the number of layers.

Rate of Advance
It would  be suitable to reduce the rate of advance to about three-quarters of the two-layer schedule to account for three layers.  And a reduction to one half of the two-layer schedule for a four-layer piece would be appropriate. The reasons for these slower rates of advance follow.

A normal rate of advance for two even layers would be about 200°C per hour to the bubble squeeze temperature.  Sometimes a very slow rate of advance is used from 50°C below the top of the bubble squeeze.  This strategy can continue to be used for thicker pieces made up of many layers with some modifications.

Multiple Layers

But for a three-layer piece, slowing the rate of advance to about 150°C is important to assist in a good bubble squeeze.  This helps get all the glass at the same temperature by the time the bubble squeeze is approached. Glass is a good insulator, and also a poor conductor of heat. This slower advance allows the bottom layer to be at the same temperature as the top piece.

For a four-layer piece, a rate of about 100°C would be suitable.  When the lower point of the bubble squeeze is reached (about 50C below the upper soak), the slow rate of advance can be used to go to the upper end of the squeeze, using the normal soak length.  

This illustrates that the more layers of glass in the stack, the slower the rate of rise must be in the bubble squeeze range.

Five Layers and Beyond

For pieces made up of more than four layers, a different strategy is needed to ensure the heat reaches the bottom layer of glass.  Graham Stone* calls this the “catch-up” schedule. It is essentially an overnight schedule with temperature equalisation soaks of 20 minutes at 125°C intervals all the way to the bubble squeeze. At each stage the rate is increased by 10°C.

This means that with a first segment rise of 20°C per hour, the second from 125°C to 250°C is at 30°C with a 20 minute soak, then 40°C to 375°C soak for 20 minutes, 50°C to 500°C and soak for 20 minutes, and finally 60°C to 625 for a final 20 minutes with 70°C to your normal bubble squeeze temperature.  This will take about 17 hours before you go on to the forming temperature.

This long heat up schedule illustrates the problem of getting the heat to the bottom layers of the stack, and the need to squeeze the air from between the layers.

Thicker pieces apply more weight to press out bubbles from lower layers, but only if the lower layers are equally as hot as the top.  This requires long schedules.

An alternative approach to this bubble squeeze problem is to fuse two layer pieces of the appropriate number to achieve the thickness required.  If these are fired with good bubble squeezes there will be a minimum of bubbles.  Combining these 6mm blanks will give fewer bubbles with a proper bubble squeeze.

Another approach is to start with 6mm glass as it comes from the maker.  This is not always possible, because it is not common for 6mm fusing glass to be made in anything but clear.  It may be possible to incorporate the clear within the stack, if it is not appropriate on the bottom.  These thicker sheets have fewer bubbles proportionally than 2mm or 3mm sheets.  So there are fewer bubbles in the final piece.

Of course, placing shards of glass at the corners, or sprinkling a very thin even layer of powder between multiple sheets will also help reduce bubbles between layers, but it is the slow rate of advance to the bubble squeeze that is the important element.

*Firing Schedules for Glass; the Kiln Companion, by Graham Stone, 2000. ISBN 0-646-39733-8

Further information is available in the ebook: Low Temperature Kiln Forming.

Layups Promoting Bubbles

Intentional Bubbles

Sometimes you want bubbles. There are various ways to achieve bubble placement with certainty rather than at random.  You can use a variety of bubble powders.  There are a variety such as the UGC bubble powder – now supplemented with bubble enamels.  The use of copper oxide powder will give bubbles of varying sizes dependent upon the amount deposited. You can also use baking soda – calcium carbonate - in the same way for clear bubbles.

You can create a range of bubble textures by arranging textured glasses in various orientations.  Fine reeded glass at right angles will give a regular pattern of small bubbles.  Accordion glass will give a slightly different arrangement.  Using fluted glass at 60 degrees to one another will give you diamond shaped bubbles if you control the temperature and time.  The variety is limited only by the textures and the way you arrange the glass orientations.

Incidental Bubbles

Most inclusions – metal, mica, organic, etc. – result in bubbles to a greater or lesser extent around the objects included.  Extended bubble squeezes are required in conjunction with a sprinkling of powder or very fine frit between the inclusion and the edge of the piece.  Sometimes corner pieces can be included in the design to keep the edges open longer allowing more air to escape.

Unwanted Bubbles

These bubbles largely come from the way in which the glass is arranged. 

Single layers at full fuse will draw in at the edges and thin from the interior, allowing any air to push up and sometimes through the glass.  This is because the thicker and heavier edges resist the movement of the air from under the glass.  This resistance, added to the thinning of the interior leads to bubbles, unless the glass is fired at fire polish or lower temperatures.

This example from Danna Worley shows the effects of firing single layers

Single layers with borders compound the problems of single layers.  The borders ensure that the edges are heavier than the interior and seal air at an even earlier stage of the firing.  The bubbles will appear between the other tack fused pieces in the interior of the piece.  Again, with this kind of lay-up, the top temperature should be no more than a rounded tack fuse.

Heavy or thick borders on two-layer bases are also circumstances where bubbles can be produced.  The border on even two-layer pieces can trap air both under the whole piece and in between layers in the same way a border can on a single layer piece.  In a lay-up like this, it is best to fuse the two base layers together first and then add the decorative pieces and border in a second firing.

This example from Andy Bennett shows how, even when inducing bubbles, things can get out of hand. Here the bubbles between layers have even thinned out the bottom layer to holes to the shelf.

Encased glass pieces are a certain way to get bubbles.  If you place even a single layer of glass pieces in a pattern around the base and then cap it with a sheet of clear, bubbles will form.  This will happen even if there are clear path ways for the air to be released from the interior.  The capping glass will not conform completely to the encased glass pieces by the time the edge is sealed, no matter how long your bubble squeeze may be.  The way to avoid this is by putting the glass pieces on top of a two-layer base.  And it is better to fuse the base layer first before adding the surface glass pieces, so they do not press down unequally, leaving a thin film of air around the heavier pieces on top.

Avoidance of unwanted bubbles

There are a few ways to avoid bubbles that are not where you want them.

• ·        Avoid using single layers with pieces on top.
• ·        When using single layers fire with slow rates of advance at low as possible temperatures with a short soak at top temperature. You will need to peek at intervals to observe when the work is finished and advance to the next segment.
• ·        Non-glass inclusions should be encased with care.  They should be as flat as possible before capped.  The bubble squeeze should be long – possibly as slow as 25°C per hour between 600°C and 677°C. This is to allow the glass at the centre to settle, pushing air from the centre out. Including a sprinkle of powder or very fine frit may help reduce bubble formation, as might chads at the corners or edge of the piece.
• ·        Organic inclusions will produce large bubbles from the combustion gases.  Use a three to four-hour soak at about 540°C to allow the burnout of the organic material before proceeding to the bubble squeeze.
• ·        Avoid borders on top of the glass.  The additional weight acts to seal the glass to the shelf and between layers, leaving air underneath to rise and even break through.
• ·        Do not cap/encase glass pieces unless you have a very good reason.  The glass pieces placed on top will stick to the surface with less chance of bubble creation, and will become flat at a full fuse.
• ·        If you must have a border or encased glass pieces, consider flip and fire – fire the piece upside down to a rounded tack fuse at least, clean thoroughly, then cap the piece and fire right side up. This can reduce the bubble formation.

Odd Schedules

Schedules appear on the internet which do not seem to have a logical sequence in the firing schedule.  Some have multiple soaks at intervals up to 540°C.  Others have kind of dance toward the top temperature – slow, quick, slow.  Some initially cool at a given rate and then slow to about half that initial rate.

Multiple soaks

These schedules have been referred to as catch-up schedules.  They tend to look something like this:

200°C to 150°C for 20 minutes

250°C to 300°C for 20 minutes

300°C to 590°C for 20 minutes

50°C   to 677°C for 30 minutes

330°C to 804°C for 10 minutes

AFAP   to 482°C for 60 minutes

60°C   to 370°C for  0 minutes

Off

The justification for the first two soaks is given as allowing the glass to catch up to the air temperature.  It would be much safer for the glass to have a moderate steady advance in temperature rather than risking the heat shocking of the glass.  You could achieve the same work in the same amount of time by altering the rate of advance to a single one of 198°C to 590°C.  This achieves the same temperature in the same amount of time, but has less risk of heat shock, as there is a steady input of heat.  

Secondly, if the glass can survive the initial rate of heat up without breaking, there is no need to soak at an arbitrary temperature.  The first relevant point where a change in temperature makes sense is above the softening point, which for most fusing glasses is about 540°C. The equivalent softening point for float glass is about 700°C

Slow, quick, slow

This kind of schedule alters rates up and down with little justification as far as I can see.  This is an example:

139°C  to 560°C  for 30 minutes

222°C  to 621°C  for 30 minutes

139°C  to 786°C  for 15 minutes

9999 to  515°C  for 120 minutes  

60°C   to 427°C  for 10 minutes

115°C  to 350°C  for 10  minutes

The question for me is why the slow down to top temperature. There is a lot of heat work being put into the glass, so that the higher top temperature may not be required.  The slower rate from 621°C does allow a form of a bubble squeeze to occur, but is not the traditional one.  A 139°C rate from 621°C to 677°C with a soak would be faster than usual, but may be acceptable.  I would prefer 50°C per hour with a 30-minute soak at the end.  Then advancing at 300°C per hour to top temperature.  The anneal soak and cool of this schedule are acceptable, even though different than I would do it.

Erratic Slumping Schedule

The fusing schedule above was followed by this slumping schedule:

83°C to 148°C  15 minutes

167°C to 590°C  10 minutes

83°C to 720°C  10 minutes

222°C to 410°C  120 minutes

83°C to 427°C  10 minutes

This schedule seems to have a catch-up phase in that it goes at half speed for the first 148°C and then doubles the speed to 590°C (a little above the brittle phase of the glass).  It then slows the rate and continues that to a very high slump temperature.  It is, of course, necessary to have a slower rate of advance in the slumping than the fusing, as the piece is now thicker. Slowing the rate of advance as much as in this should be able to achieve the slump at around 620°C (100°C) less than the target temperature used by the schedule. 

Once the top temperature soak is finished, a very slow cool to the annealing soak is used in this schedule.  This is not ideal as it invites devitrification to form.  The kiln and its contents should be allowed to cool as quickly as possible to the temperature equalisation soak at the annealing point.

The schedule then uses an annealing soak temperature 100°C below that used for the fusing. This does not make sense. The annealing soak should be at the same temperature for both firings.  The length of the soak is not in question, but the early turn off the kiln at 427°C is questionable. The anneal cool of the fused piece extended down to 350°C.  The anneal cool on slumping should be almost the same as the fuse.  Almost all anneal cools extend to 370°C at least.

Anneal Cools

Some anneal cools have erratic rather than progressive cooling.  In this example the early part of the schedule is eliminated:

……………..

AFAP to 482°C 120 minutes

110°C to 427°C 0 minutes

55°C to 370°C 0 minutes

200°C to 100°C 0 minutes

off

Here the schedule is faster in the most critical part of the anneal cool than in the next, cooler part.  This will not provide as good an anneal as if the first two segments after the equalisation soak were reversed.  Start slowly in the anneal cool and then you can speed up (approximately twice the previous segment rate) on each of the following segments.

Rationale

This critique of the schedules above is not to batter anyone.  It is to make clear that there needs to be a conscious rationale for each of the segments in relation to the others.  If you take a schedule from a source, it is a good idea to see if there is a reason for each segment and how it relates to the next. 

·        The scheduling must take account of the nature of the glass.  Glass is a poor conductor of heat and needs steady moderate input of heat.

·        Glass is brittle until approximately 55°C above the annealing temperature when you can accelerate the rate of advance.

·        Time is required to allow air out from between the layers of glass. This usually done in the range of 620°C to 675°C and is known as the bubble squeeze.

·        You need to go relatively quickly through the devitrification range of temperatures – approximately 700°C to 760°C - both up and down.

·        Glass needs a temperature equalisation soak at the annealing point (or nearby) related to its thickness.

·        The rate of cooling needs to be progressive.  The first 55°C below the annealing soak is the most important.

·        Cooling rates must be related to thickness.

·        The second cooling rate can be up to double the initial one.

·        The final cooling rate can be double the previous one.

·        The rate of firing will affect the required top temperature.

Bas Relief Moulds

Bas relief moulds that have an image carved into the surface are popular at the moment. They are most often called texture moulds.  The image is “carved” into the back of the glass, creating uneven thicknesses of glass that refract the light to show the image through the smooth plane of the front.

One of the problems with these kinds of moulds is that lots of bubbles are created, often very large ones.  This results from the many places where the air cannot escape from under the glass during the forming process.

Solutions

There are some strategies that can help avoid these bubbles.

Use the 6mm rule

Fuse the glass into a six-millimetre thickness first.  Two layers of glass give more weight to help the glass conform to the texture of the mould.  It also resists bubble formation more than a single layer.

Use the Low and Slow approach

It more important to have low and long bubble squeezes.  The most successful strategy will have a slow rise in temperature to put as much heat work into the glass as you can before the bubble squeeze.  The bubble squeeze is the most important part of firing these texture moulds.  It will start at about 600°C rising at only about 25°C/hr to around 680°C – that is, taking three to four hours. 

Use slow rates of advance

A third element is to rise slowly toward the forming temperature.  Possibly nothing faster than 75°C.  This enables you to keep the forming temperature much lower than a fast rise will.  The usual temperature recommended is about 780°C.

By using a slow rate of advance you can probably reduce the forming temperature by about 20°C.  You will need to peek at intervals to be sure the glass has taken up the required texture. Again, it is about putting as much heat into the glass at as low a temperature as possible.

Use Long soaks

An alternative to the slow rate of advance is to use a long soak at as low temperature as seems suitable.  You will need to peek at intervals to determine when the texture is achieved.  When the appropriate texture is imparted to the glass, you need to advance to the next segment.  This means that you need to know how to get your controller to skip the following segment.  Or, if the texture is not achieved before the end of the scheduled soak, how to extend the soak time.  If you are using 760°C as you target temperature with a rise of 150°C, you may wish to soak for about an hour or more.  Remember that this is in the devitrification range.

Alternative - Frit

A completely different approach is to use fine frit and powder to give a patè de verre appearance by sintering the frit.  This eliminates the bubble problem entirely.

You will need a lot of frit if you are trying to make a sheet of 6mm from the frit.  You could just take the sheets of glass cut to the size of the mould and smash them up to get the required amount of glass.  Or you can use your cullet, by weighing and smashing up enough glass. 

The calculations for weight are best done in the metric system (in cm) as there are easy conversions between volume and weight.  Assume your mould is 20cm square.  The area is 400cm².  The volume is that times 0.6cm or 240cm³.  The specific gravity of glass is approximately 2.5, so you multiply the volume by that and get 600gms of glass required to get a 6mm thick sheet. 

You could full fuse this into a clear sheet, although this would take a much higher temperature and longer soak that would be good for the mould. Better is to sinter the glass.

To sinter the glass, you need slow rises in temperature and long soaks.  A rise of about 75°C to the softening point of the glass (around 600°C) followed by a very slow rise (ca. 25°C per hour) to about 660°C is needed to allow the small grains of glass to settle together.   At the upper end of the bubble squeeze you need a three- to four-hour soak to sinter the glass. The thicker the layer of glass frit, the longer soak needed to ensure all the particles are heated.  The densest glass will be formed by a 50/50 combination of powder and fine frit.

Much better is to have a much thinner sheet formed from the frit.  This will be about two to three millimetres thick.  The weight of powder and or frit can be determined by the formula above, substituting 0.2 or 0.3 for the thickness.  This frit mixture needs to be evenly spread over the mould, with as much on the high points of the mould as the low ones.

If the mould has a lot of variation in height, you can sinter the frit mixture as a flat sheet first.  Then place it over the texture mould and give it a slow rate of advance to the to the top end of the bubble squeeze and soak for an hour or more, as required.  This will ensure you get the same thickness across the whole piece even though there differences in height.

The resulting piece will be very light and translucent.  It will have a fine granular feel to the touch.  It will have the same shape on both sides of the piece, with the upper surface having a slightly more shiny appearance than the bottom.

Further information is available in the ebook Low Temperature Kiln Forming.

Hot Spots in the Kiln

You may suspect you have hot spots in your kiln because of bubbles or one side of the piece being more fully fused than another. A good method for determining the temperature distribution across the kiln is given on the Bullseye site.  It does not require any sophisticated equipment – just supports equal distances apart and strips of glass equally wide and long – to be witnesses for the hotter and cooler parts of the kiln.  You fire slowly to a very low slump temperature – ca. 620C - for only 5 minutes.  Go as fast as possible to the annealing point and soak for 15mins. Then you can turn the kiln off, and let it cool as fast as the kiln can.

This test will show where the hotter areas are.  You will see from the test results that there is a gradual change of temperature across the shelf, rather than small hot areas that would be required for localised large bubbles originating from under the glass.  It will tell you where the cooler areas are, so you can avoid placing pieces in that area when you need precise profiles on the finished piece.

There is little to no relation between hotter areas of the kiln and localised bubbles.  Do not think hot spots are the cause of large bubbles.

Bubbles more often relate to:

Bubble squeeze

Alack of an adequate bubble squeeze, as described here.

Avoidance is the best approach.

Do not be lead into the idea that mistakes are automatically art, or that all of them can be rescued.

Rapid firing rates

Firing rates need to be adjusted to the materials you are firing.

As fast as possible firing rates can cause problems.

High temperature rapid firings can also cause problems.

Rapid firings are more likely to harm the glass than the kiln.

Damaged shelves

Distortions or damage to shelves can trap air and so cause bubbles to form between the shelf and the bottom of the glass.

Checking for flat shelves and fixing.

There are some remedies.

Volume control

Varying volumes within the piece can give problems.

There are a variety of related things that can cause large bubbles.

Glues

Glues and adhesives have a variety of effects and dangers, especially if generous amounts are used:

There are a variety of glues each with their own characteristics.

Uneven layers/layup

You must think of ways for the air to escape from the interior of the glass and from under the glass.  Most often we set up things in a way that creates bubbles. There are two main ways that we do this.

Encased items

When we put glass or other materials between an upper and lower sheet of glass we are creating conditions for bubbles to form.  The encased items hold the upper glass above the lower glass by an amount related to the thickness of the inclusion.  Routes for the air to escape must be planned. 

One of the ways to reduce the height of the space taken up by the enclosures, is to fire upside down with the inclusions on the shelf. This allows the glass -which will be the bottom layer - to form around the materials, reducing the air space between the bottom and capping layer.  This is known as flip and fire.

You then clean the face which will be capped very thoroughly.  Place the capped piece on fiber paper – which can have Thinfire placed over it, or coat with kiln wash.  This is to allow the air in the uneven bottom surface to escape from underneath through the fibre paper.

Weight

Even when there is no encased material, the weight of the glass pieces on top can create areas where the air can be trapped.  On a single layer the arrangement of pieces can create areas where the glass cannot resist the air pressure that cannot disperse from the pockets caused by the glass on top.  Very clear and generous exits for the air are required.

This can happen with two layers as well, although usually a higher temperature is required.  A means of avoiding large bubbles when there is glass – powders, frits or pieces of glass – placed on top is a two-stage firing of the piece.  First fire the base layers together at full fuse so they become one whole.  Then add the decorative elements on top and fire.  Remember to fire more slowly than for two unfired layers.  The main piece is now 6mm thick and needs a slower rise in temperature.  The additional heat work this entails may mean that a lower top temperature, or a shorter soak will be required than normal.  You will need to peek at intervals to check on the progress of the firing.

There is a multiplicity of ways that bubbles large and small can be created.  Careful layups, bubble squeezes, slower rates of advance and lower top temperatures can minimise, but not always eliminate, bubbles.

Copper inclusions

Inclusions of metals can be achieved with care.  Copper is a very good metal, as it is soft, even though its expansion characteristics are very different from glass.  This note provides some things you might consider when planning to include copper in your fused pieces.

The copper sheet should be stiff, but not thick. If the metal can be incised with a scribe and maintain that through gentle burnishing, it is suitably thick. The usual problem is that the copper is too thick rather than too thin.  Copper leaf can be very faint if a single layer is used.  Placing several layers of leaf improves the colour, but often provides wrinkles.  In summary, the requirement is to get a thickness of copper that will retain its structure, but not be so thick and stiff as to hold the glass up during the fusing process.  

Do not use the copper foil as used for stained glass applications. The adhesive backing produces a black colour from the adhesive and many bubbles -  sometimes a single large one.

Copper can provide several colours.

Copper sheet normally turns burgundy colour when oxidised.  This means that there is enough air reaching the copper to oxidise it to deep copper red.  This most normally happens, because a lot of air can contact the metal during the extensive bubble squeeze usually given to inclusions.

To keep the copper colour, clean the metal well metal well with steel wool or a pot scrubber. If you use steel wool, wash and polish dry the metal before fusing.  Reduction of air contact with the metal helps to retain the copper colour.  There are two methods I have used.  Addition of a glass flux like borax or other devitrification spray will help prevent the air getting to the surface.  Another method of avoiding oxidisation, is to cover the copper with clear powdered frit, as well as the surrounding glass.

In certain circumstances you can get the blue green verdigris typical of copper in the environment.  This is an extent of oxidisation that is between the clean copper coloured metal and the burgundy colour of extensive oxidisation.  The key seems to be be a combination of restricted air supply, shorter bubble squeezes and lower temperatures.  Experimentation is required to achieve this consistently.


The spaces under and over the copper give the opportunity for bubbles to form. 

This means that the copper needs to be as flat as possible for one thing.  Burnishing the copper can have a good effect on reducing the undulations in the copper.  Thinner copper is easier to make flat than thicker.  If you can stamp a shape from the copper with a stamper designed for card making, it is a good indication that it will burnish flat.  Thicker copper sheet holds the glass up long enough in the temperature rise during the bubble squeeze to retain air around the metal.  This remains the case even after burnishing to be as flat as possible.

The second element that can help to reduce bubbles around the copper is to sprinkle clear powder over the copper sheet once in place on the glass.  The spread of the powder over the glass assists in giving places for the air between layers to escape.

These two things combined with a long slow squeeze can reduce the amount of bubbles you get.  It cannot totally eliminate them.

Of course, a longer bubble squeeze allows air to be in contact with the copper and promotes the change to a blue green or burgundy colour.

Reducing a Bubble

A query about reducing a bubble appeared on the internet recently.  The bubble was from between the shelf and the single layer glass.  It was a relatively shallow dome that did not seem to have thinned the glass much.

There is quite a bit of information on reducing the incidence of bubbles. Among them are my blog posts on the subject.

Eliminating bubbles 
Preventing Bubbles  

Avoidance   

My view is that large thin bubbles cannot be repaired successfully.  As the bubble forms and grows, it pushes a proportion of glass to the side.  This thickens the glass at the edge of the bubble.  Bursting the bubble and filling it with something (e.g., a piece of glass, or frit) leaves marks at the thickened edge of the bubble, so it remains a mark in the finished piece.

Method 1

However, glass with a low uprising between the shelf and the glass can be successfully repaired, if the uprising is low and the glass has not thinned. In the case mentioned, the risk in simply re-firing right side up is that the bubble will increase in size. The weight of the glass may not be sufficient to pull it down except at higher temperatures – which is where the risk of increasing the size of the bubble occurs.

Instead, flip the piece over. Allow the weight of the glass to flatten the uprising. You can use a much lower temperature to flatten the glass by taking advantage of the weight of the whole piece.  This lower temperature means that you will not mark the surface so much as at higher temperatures. Don’t worry if the uprising is not central, you do not have to balance the glass on the point of the bubble for this process to work.

Take the piece to 620°C maximum for as long as it takes to flatten. The rate of advance should be slow – not more than 100°C per hour.  This steady, slow input of temperature will allow the glass to relax at lower temperatures than rapid increases. 

You should use the smoothest separator surface that you can – Thinfire or Papyros, or a smoothed kiln wash.  This together with a low temperature will give minimum markings. 

You must observe the process from about 560°C to be able to stop the slump when the piece is flat and advance to the annealing segment of the firing.

Method 2

This post gives a further alternative. Use two shelves to compress the uprising flat. Although the post is talking about thinning a pot melt, the principle is the same.  Place fibre paper around the edge equal to the thickness of the glass piece and place another prepared kiln shelf on top. You do not need to invert the surface of the piece to do this.  It may be that you will need a fire polish to remove any marks on top.

A plea

Do not drill holes. Especially not in the case of a shallow bubble.  The glass has not significantly thinned and so can be rescued.  Drilling a hole will only leave an unwanted mark.