knot Examples of Knots in Mosaics knot

These type of interleaved knots are usually called Celtic knots. However, when I looked at the book Mosaics of the Greek and Roman World by Katherine M. D. Dunbabin , there are several Celtic knots in the geometric patterns in these mosaics. I think (although I may be wrong) that all knots produced by Celts date after the Roman period, so perhaps they were copying Roman mosaics. In particular, we associate Celtic knots with British and Irish stone crosses and illuminated manuscripts, and British Roman mosaics often used geometric designs. Also, these mosaic designs frequently have outlined, multiple strands, similar to some of the Manx crosses. Remember that both illuminated manuscripts and crosses were religious, and Christianity was spread via the Roman empire, even if the Celtic church and the Roman Catholic church were later opposed.

These patterns are not necessarily to scale. Also, I'm often working from black and white photos, so the colours won't be correct. However, at the bottom of this page, there are some photos from the mosaics at Fishbourne Palace, near Chichester, Sussex (England), which show colour.





Twisted rope patterns


knot
knot
From Woodchester (England)
The twisted rope is the simplest knot of all. In fact perhaps it's too easy to be called a knot, yet it may be the first under-and-over pattern. It is common in Roman mosaics. This version is one of four corners of a frame but there are curved versions as well. The Romans seemed to like shading their knots, yet this can lead to trouble. Note that at the ends, the shading suddenly changes as it passes under a string. You can't use shading on stonework of course, but the illuminated manuscripts seem to avoid shading, presumably for this reason. The Romans seem to care less about whether the pattern was logically consistant. They were more interested in it looking good.

knot knot The twisted rope is quite easy to make a junction - from Hinton St Mary's (England)- now in the British Museum. Some of the mosaics don't have such elegant divisions - in some there is a slight air of desperation to get the string in the direction they want it!
knot knot Cross-overs are also easy. They happen in some mosaics, although others have one rope merely going underneath the other completely.

knot
knot
It's also easy for two knots to combine in a four-way plait and divide again (see right). Far left shows a crossing, although the original is more graceful, as the twisted ropes curve away as parts of two circles. This example also from Hinton St Mary's (England).




Simple plait patterns (3 way)

knotThis is a simple plait, with three strings.

knot knot From Verulamium (St Albans, England)
Here two 3-way plaits cross over. It looks very convincing, but if you look closely at the cross-over, you will discover that some of the strings disappear. The coloured version explains what is happening. You could make the grey and green strings into one string, but that would leave the centre rather empty. The shading looks more convincing, but in the whole mosaic, the illumination looks as if it comes from the four corners of the room. That might seem strange to modern eyes, which are used to a single light source in the middle of the room, but might make sense to a Roman who would have oil lamps lower down, possibly in the corners.
knot So how can we cross over plaits without losing strings? First, lets remove the surplus string (see left). This leaves a gap in the centre, and also we've lost the over-and-over alternation. If we tidy this up, we get one of a number of solutions to the cross-over problem. However, these all tend to have problems. The orientation of the plaits get altered by the cross-over, and the cross-over becomes staggered, so plaits going into the cross-over are not aligned with those coming out.
knot knot knot knot
Two go off to the left and one to the right (based on above) One goes to the left, one to the right, and one straight on All three go straight on All three go straight on, but less stagger

So we can see why the Romans 'cheated' a solution. Four way plaits have no problem at all (see below). The illuminated manuscript illustrators seem to prefer more complex plaits than 3-way, which gives them a greater choice of patterns. They may have had greater freedom to make any pattern they wanted. The mosaic makers seemed to use specific patterns in required ways, which may have led to impossible requirements, such as the cross-over of 3-way plaits above.



knot knot From Verulamium (St Albans, England)
This is from the same mosaic. Here a 3-way plaits splits into two. Again, it looks convincing, but one string disappears again. The problem is worse than the cross-over (above). There are solutions to the cross-over, if inelegant, but there is no solution to this one. There are an odd number of strings going into or coming out of the junction (3 x 3, making 9). If we expect a string going into the junction to also come out again, then this takes up 2 strings of the total. So four strings came go into the junction and come out again, but that leaves one going in and staying there. If you look closely, you will see that it's the yellow one in my coloured version.



4 way plait patterns

From Verulamium (St Albans, England)
This border is a simple four way plait. The regular pattern of dots looks as if it may have been designed using the dots method. The colouring shows two shades, but the middle pieces have more light and less dark, while the outer pieces have more dark. The second colouring shows how it can be made with two strings.

knot
knot
knot This 4-way plait with corner comes from the House of the Triumph of Neptune, Acholla, in North Africa.
knot We've seen above that simple plait cross-overs have problems, and junctions are impossible. However, if they changed the problem slightly then the solution is easy. If you change the plaits to 4-way, you can cross-over without a stagger, and preserving the orientation. This does happen in mosaics, such as in the villa at Piazza Armerina in Sicily.
The four-way plait has no problems with junctions at all. Since there are an even number of strings coming into the junction in each plait, there are no spare strings to get lost.
knot



5-way plait patterns

From Hammath Tiberias synagogue, in Palestine.
A corner with a 5-way plait. There seems to be no enthusiasm for more complicated patterns with 5-way plaits, probably sensible considering the problems they had with 3-way plaits.

knot




Simple knots

knot knot This very simple knot is at the centre of a mosaic from Woodchester (England).

knot Here is a more complicated one from the House of the Nypmphaeum at Ostia. It's basically the same knot, but with six strands. However, the strands are treated as one unit, going over and under together, unlike some illuminated manuscripts, where they behave independently of each other. This didn't seem to have occurred to the Romans.


knot This knot is coloured in the mosaic to show the different strands (not necessarily with these colours, since I am working from a black and white photo). It starts as double strands working as one in the centre, which go their separate ways at the edge.




Other patterns


From the church of St George at Deir el-'Adas, south of Damascus, Syria

knot
knot
This border is interesting as it is based on reef knots. Generally Celtic knots use strictly over and under patterns, which, perhaps surprisingly, real knots don't do. In fact, you couldn't tie this border with real string or rope, as it is made up of loops, as shown on the second colouring. However, if you had some loops, you could make this pattern, since the loops are not connected with each other. Although they look as if they are interleaved, you could pull each loop out, one at a time.




From the Yakto complex at Antioch
knot
knot
This pattern reminds me a little of an Isle of Man pattern. It seems to be a set of interlocking loops, but there is a covered-over piece (which is missing in the Manx pattern). This means you can interprete the pattern in two ways - see right. I think the colouring suggests the top one, but then you have loose ends tucked away begind the covered bit. Perhaps that's why it's there! It could be just a visual pattern with no 3D equivalent. knot
knot
There is a similar pattern at the villa of the Falconer, Argos.
knot This can also be interpreted in two ways, either as two twisted ropes with unexplained vertical bars, or as interlocking loops. knot knot

knot This is an interesting knot from Thebes, which is angular rather than curved.


Fishbourne mosaics

Fishbourne Palace, near Chichester, Sussex (England) is a grand Roman villa with splendid mosaics. Here are some of the Celtic knots on them. There are also msaics with strange black and white patterns, which shift perspective as you look at them. Click here for photos of those.

Fishbourne mosaic with Celtic knots Fishbourne mosaic with Celtic knots Fishbourne mosaic with Celtic knots Fishbourne mosaic with Celtic knots





You may be interested in my interactive webpage where you can design your own mosaic.

Return to Knots index.

© Jo Edkins 2003