The Interior of My Brain: A Knitting and Fiber Arts Blog

I have been using modular machine knitting techniques for some time as a way to create design elements while extending the width limitations imposed by a 200-needle knitting machine bed. Until now I have avoided concocting designs that had half of the neckline knitted vertically and the other half knitted horizontally because that requires one set of shaping maneuvers for the side-to-side knitting and another set of maneuvers to shape the up-and-down knitting. This is necessary because knit stitches are wider than they are tall, at a ratio of three stitches to four rows. I might have used the 3:4 ratio on my own to figure out the formula to get the same curve on either side, but I’m busy/lazy, and surely someone else must have already invented that wheel.

And someone had, an Austrian designer who goes by the name Jutta von Hinterm Stein. Hinterm Stein translates as “behind the stone”, so I assume that’s probably not Jutta’s last name. No matter, last spring she published a pattern called Dinner with My Captain that did exactly what I’ve been wanting to do with directional knitting. The pattern is on Ravelry, but if you’re looking at her website, it’s the blue and white striped buttoned shirt with one half of the striping going horizontally and the other half vertically. So I bought the pattern, thinking I would borrow the numbers and knit it up on the machine. But the pattern was much more detailed and complicated than I was expecting, with techniques I had never seen before, and I’m not great at understanding patterns under any circumstances until I actually follow them step by step, stitch by stitch. I decided I would hand knit my first version of the garment so that I could internalize the ideas before I tried to adapt them to the machine.

Yes, it was a whole lot of stockinette, and I don’t like knitting plain stockinette, but it wasn’t going to be brainless stockinette because of the striping pattern and all the shaping and edge details. Also the yarn I was going to use was four colors of a very thin hemp yarn by Lucci that Melissa gave me, after a customer’s widower gave her his late wife’s stash. I decided to use the blue yarn as the color of the edgings, but aside from that I was going to try to randomize the use of the four colors, while carrying all the colors along the edge, as the pattern recommended for the two colors, and try to avoid repeating the color sequences as much as possible. That would help keep me awake. I wanted my shirt to be very light and airy, because summer is hell and most of my summer hand-knits are too warm in the worst of it even when they’re loose and made of plant fibers. I made a swatch using a U.S. size 4 needle, which was large for the thinness of the yarn but gave me the airy fabric I was looking for. I washed the swatch and got a gauge larger than Jutta’s, so I estimated that the numbers for her fourth size would give me the unencumbered fit that I wanted from this shirt.

But I had never knitted with a hemp yarn before and was unfamiliar with how it would behave, and this uncertainty plagued me throughout the entire process. The first thing I did wrong was to use a too-large needle for the button band, which was the first step of the process, although I did go down one size. I should have gone down three sizes. The too-large needle gave me a floppy, uneven, unsubstantial band that hung lower than the rest of the piece, which alarmed me by stretching lengthwise so much that I doubted I would get the width I wanted. I knew pretty early on that I was going to have to unpick the button band and reknit it at the end on U.S. size 0 needles, with better buttonholes. I decided to use the numbers for the sixth size for the width, and throughout the process of constructing the four modular pieces, I toggled between numbers for the fourth and sixth sizes. I tried to be faithful to the numbers for the fourth size when it came to the length of the pieces, but when I got confused, I defaulted to the 3:4 ratio for connecting horizontal pieces to vertical pieces and the inverse for connecting vertical to horizontal. As I worked, I was distressed by the fact that the horizontal pieces were much, much wider and shorter than the vertical pieces, due to the looseness of the thin hemp yarn knitted on needles that were large for the yarn.

The pattern had a method for connecting the side pieces that I had never seen before, executed by keeping the stitches live on the vertical piece, picking up the same number of stitches on the horizontal piece, putting the stitches from each side onto one needle in alternating order, and finally slipping the stitches over each other one by one and off the needle. It produces a neat and elegant join. It’s also a ton of work. Also in my case, the discrepancy between the row gauge and the stitch gauge produced a ripple along the join that I fervently hoped would come out in the wash. I wish I had taken pictures of how awful the unfinished work looked, but I was too embarrassed to document it. These were the days that challenged my faith in the 3:4 ratio and the power of blocking.

But on I slogged for two and a half months of tedious stockinette with bits of garter stitch, at least as tedious, to relieve the stockinette. But I really wanted to see how the placement of the increases that shaped the neck would work out in real life, since I had seen some finished garments whose necklines looked good and some that looked awful. I admired the way the pattern avoided cutting the yarns and creating ends, but I really added something to carrying the yarns along the edge by using four of them, which I tried to use in a way that seemed random. At least it helped keep me awake.

When I finished the body pieces, with the second button band knitted on a U.S. size 0 that made a sufficiently solid fabric for the buttons to sit on, I had to unpick the button band that started the piece and reknit it. The pattern calls for a very small yarnover buttonhole, but my buttons were too big big for that, and I thought it looked flimsy and sloppy anyway, so I improvised my own buttonhole when I didn’t have the bandwidth to follow the instructions for the Tulips Buttonhole. I wrapped my working yarn around the base of the stitch before the hole a couple of times, then wrapped the yarn around the base of the stitch before it in a figure-8 and left the yarn at the front of the work. Then I picked up one of the wraps around the first stitch and passed it over the first stitch of the hole, and passed that stitch over the next stitch until I had an opening that was right for my buttons. Next I turned the work around and did a cable cast-on between the two stitches before the opening until I had one more stitch than the number of stitches I had bound off for the hole opening, and passed that stitch over the stitch at the end of the opening. Pull the yarn tight, wrap the yarn around the stitch a time or two, then figure-8 the yarn around the next stitch after it, and resume knitting when the yarn is in the right position to knit the first stitch after the hole.

Finally, after the neck edging was knitted, it was the moment when I would find out if I had wasted my summer of knitting. Time to block this misshaped monster. Happily, blocking performed its miracle of smoothing out the rippled joins and (mostly) restoring the 3:4 ratio, to create a loose, flowing, airy summer shirt that turns out to be the most comfortable garment I have for stifling summer weather. It’s a bit see-through and the armholes reveal a little more of the side of my body than I like, but these violations of modesty are nothing compared to the other alternative in that kind of weather, which is stripping off every stitch of clothing and wearing nothing. In the end, I was very pleased with the garment and the pattern, because the numbers produced a well-balanced neck shaping and shoulder slope and all those fussy little details gave the garment a lovely finish. My next task was to translate the design and numbers to machine knitting.

Since my first effort produced such a large garment, I decided to try the pattern’s intended fit, a size that provides of 2″ of ease. Two inches is a lot less ease than I’m usually comfortable with, but I thought I’d give it a try anyway. I knitted up a swatch of my yarn, Cascade Cotton Sox, and it seemed pretty close to gauge. Maybe a bit small, so I loosened the knob a click to 8.1 and went up a size in the pattern. This time I was going to use only two colors and follow the numbers for the chosen size as closely as I possibly could. Also I decided that the garter stitch edges were essential to the design, so I was going to do them as written even though machine-knit garter stitch, without a garter carriage, is hand-manipulated and labor-intensive. I don’t have a garter carriage and don’t know how they work. But I was prepared for the hand manipulation of using the latch tool to undo the stockinette stitches made by the movement of the carriage, one at a time, and re-form them to reverse them to knit stitches on the facing side of the work, which is always purl on a knitting machine.

Another thing machine knitting did to the interpretation of the hand knitting pattern was to reverse the directionality of the design, since the machine knits with the purl side facing and the pattern was written with the knit side facing. So the first piece, written to be the right front, became the left front in the machine-knit version, and required the buttonholes to be knitted on the final piece of the construction rather than the first piece. I didn’t want the headache of starting off the machine knitting with multiple garter ridges, and I left the button band to hand-knit it at the end. I started the machine knitting with the instructions for the vertical striping of the front body after the garter stitch button placket.

Adapting the instructions to the machine was easier in some ways than hand knitting it because the needles are numbered and the row counter does exactly that automatically, which makes it much easier to keep track of the numbers than in hand knitting. To make the best use of the row counter, I divided the pattern into sections of related maneuvers, added up the number of rows for the size in the given section, and set the counter to 0 and knitted the rows and performed the maneuvers at the specified intervals. Another way the machine knitting was easier than the hand knitting was that I could do the seaming as I knitted, by joining the edge stitches of the bind-off on the vertical knitting at the 4:3 ratio, which is one of my go-to maneuvers. It’s so easy on the knitting machine, thanks to the row counter. Pick up an edge stitch and put it onto the edge needle, knit back and forth, pick up the next edge stitch and knit back and forth, skip the next edge stitch and pick up the stitch after it, knit back and forth. It boils down to skipping an edge stitch whenever the stitch counter is on a multiple of four. The numbered needles made it easy to get the right stitch count when picking up stitches along the edge of the horizontally knitted piece, for the vertically knitted second half of the back.

Of course in machine knitting one must always think before moving the carriage, and there were a list of things that needed to be done before moving that carriage for this project. Make sure the carried yarn has been placed into the hook of the edge needle before knitting the other color. Make sure the slipped stitch edge needle is in hold position on the return carriage row before knitting it, to create a smooth edge. Then re-form the edge stitches for the garter stitch edging before knitting the next row. Keep track of the row counter to do the right number of shaping increases/decreases in the right places. Keep track of the striping pattern and change colors accordingly. Having accepted as a base line the fact that machine knitting requires the knitter to keep their wits about them at all times, it was the garter stitch that was the only thing that wasn’t easier on the machine. I did the garter stitch edging on the horizontally knitted back piece by re-forming every one of the 90 stitches in the row for each of the five ridges of the edging. That took a couple of hours, probably the same amount of time needed to do them by hand, which I did for the edging on the second front and all the other garter stitch edgings on the sleeves, button bands, and neck band. I started thinking about having another go at the garter bar, which I already knew from failed experience was not as easy as the master machine knitters in the videos make it look.

Every time I took a finished section off the machine and held the growing garment against my body, I had a creeping inkling that grew with the garment, that my gauge swatch had failed me. Instead of the generous size I thought I was making, I was really making a little toy-sized shirt that was going to fit my toy-sized sister Judy, whose birthday was right around the time I was finishing this shirt. I held out a dim hope that maybe the yarn would relax after blocking or that wide button bands would give me the width to wear it comfortably, but I had made the short size, and it was even shorter than the pattern said. No, this was not my shirt. I’ll give it to Judy at Thanksgiving. I hope she likes it.

Now I had a real-life full-sized stitch and row gauge for this pattern, thanks to this too-small effort. I also had more yarn and a color scheme and a strong desire to internalize the pattern’s shaping and construction concepts so that I can apply them in my own designs. I also wanted a version of the garment that didn’t flash glimpses of my flesh and undergarments when viewed from the side, and I needed a bit more length. So I went up two sizes and chose the longer option for the pattern. Theoretically, the size I chose was supposed to be 53″ around, but I knew that this yarn at the gauge I was knitting wouldn’t knit up to be that large. If the result was three inches larger around than the too-small version, it would be something I could wear.

Before I embarked on this third version, I studied up on the use of the garter bar. I practiced for several days until I got pretty automatic on the steps for the maneuver and was fairly consistently able to do the flip so that at least some of the plies of all of the stitches made it back onto the needles. Repairing the stitches probably took as long as re-forming them through hand manipulation, but I’m willing to accept a lot of imperfect success on my way to achieving perfect success. Finally I started knitting, and it moved along as quickly as the hand-manipulated garter stitches would permit. Then I got to the horizontally knitted first half of the back, which ends with five garter ridges across the width of the piece. Garter bar time! Well, the process wasn’t pretty or elegant, and I reconstructed hundreds of stitches before I was done, but a couple of hours later I was done and could bind off with a hearty sigh of relief. Maybe things would go better when I got to the same operation on the horizontally knitted final piece on the front. Well, they didn’t. Things went worse. I lost half my stitches on the first flip, so I gave up and put everything on a circular needle to hand knit that edge and all the rest of the of the garter stitch edges.

When I got the four pieces off the machine, I took a good look at my color scheme. I didn’t have enough of the lime green and teal blue left to make a second shirt, but I had enough to get me about halfway through, and I had a green that was pretty close to the teal and a mustard yellow that had a hue and value relationship to the lime. My idea was to use up the colors and then work in the replacement colors. Would it look cool or like a mistake? I was hoping for cool, but I was willing to accept it looking like a mistake, because I knew that if the garment was well knitted, and it was, and well fitting, which I hoped it was, the gravitational pull of everything else being right would pull the peculiarity of the color combination into the realm of ambiguity rather than outright mistake. I’m fine with ambiguity. But everyone who has seen the shirt, in photos or in person, has particularly praised the colors, which is very nice, but the thing that makes this one the Goldilocks shirt– not too big like the first shirt or too small like the second shirt– is the fact that this one is just right, with the circumference that the too-small shirt was supposed to have had.

I’m not done with this pattern, even after three versions, one hand-knit and two machine-knit. I have yarn to make shirts for my two daughters, and right now my younger daughter’s version is on the machine. After the too-small version, I don’t know why it didn’t occur to me to just loosen the stitch size knob a couple of clicks in order to get a fabric that’s closer to the pattern’s gauge, but finally I got a clue. T9.0 doesn’t completely match the pattern’s gauge of 25 st/30 rows over 10 centimeters, but at 27 st/33 rows over 10 cm, going up only one size from my daughters’ normal size will give them a pleasing fit. I have also used the numbers for a larger size to add more length to the new version, because my daughters are taller than I am, and big surprise, now that the gauge is closer to the target, the shirt actually is going to be pretty long. I’m also going to take another crack at the garter bar, because it would be so great if I could master machine-knit garter stitch and do it accurately and efficiently.

These additional two shirts will also be an opportunity to find a better way to do the neck shaping of the second front, which calls for a series of make-1 increases on successive rows. Normally, in machine knitting, I do an M1 by poking a transfer tool into the stitch next to the empty needle and lifting the loop from the row below and putting it onto the empty needle. But if you do that on successive rows, you’re lifting the same bit of yarn for increase after increase, which produces a distorted fabric. It’s really bugging me on the two machine-knit shirts, although most people wouldn’t notice. So I’m going to put the transfer tool into the space between the two stitches where I want to increase, twist the yarn between the stitches, and put the loop onto the needle. There will probably be a small hole, but at least the fabric won’t have that distorted pull disrupting the curve of the neckline. Somehow this didn’t happen in the hand-knitted version. I can’t remember what I did differently.

Now that I’m working on my fourth iteration of this pattern, I’m able to isolate the elements that I can apply to other contexts, designs, yarns, and gauges. The maneuvers for the front and back neck curves are the same for every one of the nine sizes, so I can apply that formula to whatever numbers I might come up with. In fact, why would I have to come up with very many numbers when there are nine sizes that I can plug my stitch and row gauges into? The vertical-to-horizontal ratio for the shoulder slope– a decrease every 6 rows: short row 5 stitches apart– should work for yarn weights from fingering to DK, although heavier yarn with a larger gauge might need different numbers. Also, if I ever get good enough at the garter bar flip to incorporate garter stitch edgings into my machine-knit sweaters, Jutta’s ratio of five horizontal garter ridges to six vertical ridges will be my go-to. I think that a forensic analysis of my future work will reveal traces of Dinner With My Captain in garments that don’t look anything like it.