In my last post about flattened diamonds, I talked about the problem of dealing with a diamond when the desired stitch pattern is centered around a single stitch.
The focal point of such a design is off center when placed on an even number of stitches. (I’ve been having trouble with my chart software. It’s almost fixed, but I thought I’d take this opportunity to show that graph paper can be just fine for the design process.)
The diamonds I’ve been working with so far have a rectangle in the middle based on an even number of stitches. When I place a simple chevron on such a diamond, it looks like this, which bothers my mirror symmetry-loving brain, especially when designing a shawl. The actual pattern repeat works out just fine, mind you! But there is the necessity of including an extra stitch on the right-hand side of each row.
My first thought about this was to make the center repeat an odd number of stitches, and make one arm of the diamond one stitch longer than the other. This certainly tiles, but it makes the increases at each edge peculiar.
My second thought makes use of 3-to-2 decreases and the principle that the decrease corresponding to a particular increase can be on a completely different row. (On this sketchy chart, the dark squares are grey-no-stitch squares.) This tiles nicely, keeps the increases at each edge consistent, and makes the design process easier for me. It won’t work for all stitch patterns, however. But I do like it.
So my third thought is probably too complicated for a final pattern chart, but it helps satisfy my mirror symmetry cravings in the design process. The arrow points at the spine of the shawl. The central column of diamonds has an extra column up the middle, so the chevrons can be centered on that spine. The diamonds on each side are based on an even number – the chevron sits slightly to the left in the diamonds that are on the right-hand side of the pattern, and vice versa.
Note that in the end, once the diamond outlines are removed, this produces the same overall rectangular stitch pattern as the second chart in this post: a repeat of 12 + 1 stitches and 8 rows. But if designing a crescent shawl, I think the third thought works best.
String Geekery 
Last night I got to my swatch it pile and tried your L, J and upsidedown T decreases from your (brilliant) fillet mesh. I wanted a twist stitch circle-ish pattern, so I was comparing the upsidedown T to a centered double increase. The upsidedown T was very squared (a cool effect I’d like to look into more) but it mirrored the CCI better if I worked the stitches first, then decreased them, because then they were more angled.
That makes good sense, and I’m glad you’ve hit on a solution you like.
Just wanted to say that the graph paper looked perfectly normal to me and I hadn’t even noticed that you weren’t using the digital version until you mentioned it! Which is to say, I like graph paper & always have (even before I used it for anything “math-y”)! 😉
Graph paper is awesome.