The swing experiment – knitting by spreadsheet

Experimental swing knittingSo, I’m trying to understand the short-row knitting technique that is variously known as swing or tapestry knitting.  The fabric is built up, not in regular, row-by-row lines, but in sections of a few stitches at a time that create (2-D) waves and hills.  You start each section at the right hand side of the fabric, knit across to the start or end of the “wave” (or sometimes knit right to the end of the row and then back to where the wave is to be placed), then work back and forth over a portion of the stitches on the needle for a few rows, shaping your wave as you go, before finishing with a wrong side row that ends back at the right hand side of the piece.  The troughs or valleys between waves then need to be filled in by subsequent short-row sections, or fabric distortion results.  In other words, the waves must be staggered across the fabric such that they mesh together.  If alternate sections are knitted in different colours, the result is a dramatic, curvily-patterned swatch with the short rows of knitting tilted this way and that.  It’s intarsia-like, but not intarsia, because only one colour is used at a time.

The challenge – for me, at least – is to work out how to make these short-row sections fit into the fabric while achieving the correct overall shape, whether that’s a rectangle for the front of a sweater, a triangular shawl or something more complicated still. There’s no reason why the swingy waves couldn’t also be used to provide 3-D shaping for a garment, but I think that’s a step too far at present.

It seems to me that, to keep the fabric flat, it must be important to have an equal number of rows beneath each stitch on the needle, when averaged out over a few inches of length. In particular, efforts must be made to keep the rolling average of rows at the edges close to the overall average, or the sides of the piece are likely to pull in or become wavy.

I did worry at first that the task would be more complicated than simply keeping track of the number of rows across the whole width, because the tilting of indiswing-chart-1vidual sections will mean that the effective height (and width) of a stitch will vary according to the degree of tilt.  But I managed to convince myself that the fact that the sections mesh together means that the average degree of tilt beneath any given stitch on the needle is zero.  I haven’t proved that yet, and frankly I’m not sure I can be bothered, but the swatches I’ve been knitting seem to bear it out.  It may be desirable to balance the number of right-leaning sections (like the yellow one on the right) beneath each stitch with left-leaning (red) sections, because they start and end differently, but I doubt that any difference to the degree of tilt is significant.  I’m heartened by the fact that my swatches knitted in acrylic are ending up rectangular, which means that there shouldn’t be any problem at all when knitting in wool or another blockable fibre.

Having used the spreadsheet to work out how many rows go into each section – the answer is 12 in total – and how many rows are under each stitch in that section (that’s what the numbers are in the chart above) – which is 2, 4, 6, 8 or 10 in the central zones, with up to 12 at the edges – I was then able to simplify things by using a single spreadsheet row for each section, like this:

swing-chart-2Then, all that’s needed to make the sections mesh together properly is to vary their horizontal placement and keep track of the total row count.  It turns out that both left- and right-leaning sections have the same row profile, and hence the same shape.

The experimental swatch at the top of this post, and the charts, are based on the Mixed Wave Cowl which uses evenly spaced short-row sections that alternate between leaning to the left and to the right.  It’s definitely worth reading that blog post.  Each section is centred on a particular stitch marker and worked between the markers on either side of it.  I placed my markers at 8 stitch intervals with just 4 stitches at each edge, giving 3 central sections of 16 sts (counting from the left-most stitch to the right-most) and 2 edge sections of 12 sts.  They overlap, as you can see from the swatch, which is why a total of three 16-st zones and two 12-st zones fit into 40 sts.

The numbers in each coloured section of the second chart are the number of right-side ribs at that stitch position, not counting the 2 rows (1 rib) per section that are worked across all the stitches on the needle.  In other words, if the number of rows in the previous chart is X, then the number in the chart immediately above is (X-2)/2. I just find it easier to work with the smaller numbers.  Now I can copy and paste the relevant cells from a row for a particular section – a left-leaning red one or a right-leaning yellow one – into the appropriate position to represent whatever section I am planning to knit next and sum the column beneath each stitch number (numbering from the right) to give me the total number of rows below each stitch.

With these even-width, evenly spaced sections, it’s easy to see from the chart that the rows will balance out provided that the 5 markers are used equally as the section mid-points.  In other words, every five sections knitted should include a section centred on each of the markers.  They don’t have to be worked in the same order each time, and it would probably be unwise to have only one or two other sections between two that are centred on the same marker.  Also, I think working several sections together on say the left half of the piece would be a mistake.  The aim should be to choose the next section to knit quasi-randomly rather than totally randomly.  Or you could just work them in a repeating sequence for a more ordered look, or perhaps reverse the sequence half way through to produce a mirrored design.

Although my first sample and associated spreadsheet charts are based on the Mixed Wave Cowl, they aren’t identical.  If you feel like having a go yourself, I should ignore what I’ve done until you have knitted a few sections of the cowl.  Very clear instructions are given in the pattern together with an explanation of the principles on which they are based, it’s not necessary to know anything about swing knitting in advance.  Do learn how to do German Short Rows if you don’t already have that skill, because for garter stitch short rows there’s nothing better.  The cowl does involve increasing at one edge and decreasing at the other to make a parallelogram instead of a rectangle, which adds to the fun.

I’m now wondering how to get the fabric even when placing the individual short-row sections more randomly.  I could chart every section on a spreadsheet, but that is going to be very tedious, given that a full width of say 60 sts might have 7-10 sections each with perhaps 14 or 16 rows.

About yorkshirecrafter

I live and work in West Yorkshire.  I've always enjoyed crafts of all types, from woodwork to lace-making.  I also enjoy anything mathematical, which makes knitting a favourite pastime, especially complicated designs.  I've been advising businesses and industry on environmental matters for 30 years and also have an interest in green living, especially where it saves me money. I live with my husband and our Maine Coon in a 100-year-old cottage that constantly needs something doing to it.  Fortunately, I enjoy DIY too.
This entry was posted in knitting and tagged , . Bookmark the permalink.

1 Response to The swing experiment – knitting by spreadsheet

  1. Pingback: Not the rhubarb triangle | YorkshireCrafter

Please leave a reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.