Apparently my last post caused some confusion. A couple of people wrote to say that they didn’t understand why the original graph wasn’t knit-able. I’ll try to explain again.
One principle of lacy and other texture knitting is the equivalence of decreases and increases. In patterns with parallel edges, for every new stitch introduced into a row via a YO, Make 1 or cast on, there is an equivalent stitch removed by a complementing decrease. That decrease can come in many forms – a K2tog, a SSK, a cast-off, or as part of a multiple stitch decrease (Sl-K2tog-PSSO, K3tog, SSSSK, etc.) Yes, there are some exceptions – patterns that deviate by having a decreased stitch count on one or more rows, but if they have parallel edges, they must restore the stitch count on subsequent rows. I’ve graphed both types here before.
Walker’s Porcupine Stitch uses increases and decreases balanced throughout to maintain parallel edges
By contrast, her Starlight Lace Stitch is a parallel edge insertion that has a modified stitch count on rows 16 and 14, that is restored in both places on the next right-side pattern row. The presence of those evil gray “no stitch” boxes is a dead give-away that stitch count monkeying has happened.
To create a panel with one or more decorative edges – edges that zig in and out to make nifty curves, scallops, or sawtooth or triangle points – the stitch count has to be deliberately altered so that the width of the piece grows and then shrinks in a predictable manner. Most of these decorative panels are edgings – strips with one nice firm straight edge that is usually knit or sewn onto the thing being trimmed, and one fantastical dagged edge – the decorative points or ruffles that hang free. There are two-edged edgings that in the past were used as trim or decorative strips all by themselves – lingerie straps, camouflage for shelf edges, free strips appliqued onto towels and house linen, but they’re far less common and are rarely seen in modern pattern collections.
The stitches introduced (or decreased) to form the points can occur anywhere in a row. Placement as well as the number helps determine the overall shape and depth of the point. If the new stitches accumulate or disappear from the left of the location of increase/decrease, the points tend to be a bit sharper. If they accumulate between the stable edge and the location of the increase/decrease the points formed are more like waves or scallops. We saw that in the pattern I charted in the last post, where the point-forming increases/decreases were relatively close to the stable right edge of the piece, stitches were accumulated between the stable right edge and the location of increase decrease, and that spot was followed by a relatively large section that had a stable stitch count. Here are simple graphs of a few basic edging shapes, stripped of all lace detail. Note that in each and every one, if a row has more (or fewer) stitches than the one that preceded it, there is a clearly discernible cause on that preceding row – an increase or a decrease that’s clearly to blame.
It’s the absence of any stitch-to-blame in the historical chart that made it un-knitable:
Yes, the graph looks good. The points march in and out in clearly defined order – but the causes for that patterning are absent. Every YO on this graph is countered by a decrease. There are none left over to form the basic triangle point shape.
Now as to why the chart was published this way – the pattern book I was working from is a direct facsimile of a work produced in Germany in 1921, in a language I can’t read. I did double check the instructions, both against the English key thoughtfully provided by the book’s modern editors; and against the original diagrams presented at the front of the book. Those show standard symbols and a little engraving of what the resulting work should look like. I also successfully reproduced another pattern on the page that uses the same symbols, so I’m pretty sure that in spite of not being able to read the accompanying text I didn’t miss anything substantive.
My guess is that because charting was new, and the symbols used in the original book are not standard (charting symbols aren’t standard even today), among the pattern designers, the artist that laid out the pattern, the typesetter, and the proofer, errors slipped in. Proofing knitting patterns isn’t an easy thing, as any modern professional pattern writer/editor can tell you. In my experience, the most accurate patterns appear to have been produced between 1950 and arbitrarily – 1985. Stuff before in general isn’t as stitch for stitch perfect or isn’t in modern notation; and stuff after seems to have suffered from a lack of skilled manpower and/or editorial time. Not to say everything published after 1985 is junk, but we’ve all seen books rushed to market that required dozens of pages of errata. Books published during the designated “sweet spot of knitting” era tended to require far fewer corrections than do many contemporary works. Kudos to those professional authors/editors/publishers who have taken on the extra time and expense in pursuit of perfection. Eyebrows are raised at those who cut corners. Slack is cut for pre-modern works, especially those that pioneered new forms of instruction.
So the moral of the story in knitting as in far more weighty world matters, is “trust, but verify.”
Afterword: People new to charting might find the Charting 101-107 series here on String useful. You can find those posts under my Reference Shelf tag.
Apparently my post on knitting patterns from books published prior to 1920 or so has struck a chord. I’ve gotten a couple of requests on how to go about translating these older knitting patterns to modern notation. I did a six part section on how to graph up patterns from written notation before (Charting 101, 102, 103, 104, 105 and 106 ), so this sort of follows as optional post-lesson workshop.
This time I’ll start with a web-available pattern. K. Harris at Vintage Connection has posted a transcription of a knitted insertion pattern that first appeared in The Delineator magazine, in June 1896. I’ll be producing a modern notation graph for that lace panel. Before we begin, it’s worth noting some common features of turn-of-the-century knitting. Not every technique known today was widely used, and terms varied a bit – even more widely than they do now. I’ll try to cover some of the most common notations.
Knit and purl – k, p
Not much difference. Basic knits and purls were pretty much as we know them. There were however a couple of associated usages that are less common today. Knit plain usually meant work in knit stitch only. One complication – it follows then that for things knit in the round knit plain came to mean “work in stockinette.” Occasionally by extension knit plain was used to indicate stockinette done in the flat rather than in the round, even though intervening rows of purl by necessity exist. I’ve also seen it used very infrequently to mean “continue working in established pattern,” but that’s rare. More often the term work even was used in that context.
Another alternate usage – purls were sometimes referred to by the term seam, as in the instruction “knit two, seam two” to produce k2 p2 rib. This is probably a hold-over from early sock making, in which a column of purls on the back of the leg was used in imitation of a seam line.
Narrow – n, k. 2t, t,
The modern equivalent of narrow is K2tog – the standard right leaning decrease. Sometimes this is written up as K2, with the “tog” part of K2tog being left out entirely. Older patterns did not use SSK. Occasionally they call out a SSK equivalent of “slip one, knit one, pass slip stitch over” (see below) but most often they don’t bother with a left leaning decrease, and use K2tog, even when the cognate would be visually more balanced or appealing. Close inspection of accompanying illustrations reveals that the knitters did employ K2tog for almost all decreases. Less frequently this decrease is referred to as together (t) or knit 2 together (k. 2t.).
One unusual notation on narrow – a couple of patterns I’ve seen use n followed by the note “by slipping the needle through the back of the stitches.” This does sound a bit like a proto-SSK. But unless otherwise modified or explained, it’s pretty safe to assume that any n means k2tog.
Slip – s
Another movement that’s pretty standard. Unless otherwise modified, slip in historical context means slip purlwise – transferring the stitch from the left to the right hand needle without changing its orientation.
Slip and bind off – sb, sbo, sl&b,
Another historical way of referring to the left leaning decrease or SSK equivalent, this refers to the s1-k1-psso unit.
Over – o, th, w, tho, th. o,
Yarn overs or eyelet producing increases – still a source of multiple terms today – have even more names if you go back through time. I’ve seen YO referred to as over (o), throw (th), throw over needle (tho or thn) eyelet (e), widen (w), make (m), put over (po), yarn on needle aka yarn over needle (yon), wool round needle (wo, wrn, won).
Special note on double YOs. Most of the time modern patterns use a multiple-unit YO if a really big eyelet is needed. But in historical patterns when YOs were used to make columns of fagot-stitch lace, it was common for the YO that formed them to be specified as a double yarn over, probably because of the yarn manipulation used to create them needed to allow for a subsequent p2tog. If a pattern with fagoting calls for a double yarn over but the stitch count on the subsequent row doesn’t account for the additional new stitch (or doesn’t mention dropping it), it’s a good indication that a modern redaction will call for only one YO and not two.
Make – m
This can be problematic. It’s on the previous list as a euphemism for YO, but it is also used in historical patterns for invisible increases – where an additional stitch is added without creating the eyelet hole formed by a YO. Modern “make” is usually interpreted as a raised bar increase, although other forms of adding a stitch like knitting into a stitch on the row below are also sometimes used. A bit of close examination of any illustrations or even experimentation may be called for here. The term made stitch is also sometimes used to indicate the new stitch formed by a YO in a previous row – especially when more than one YO created multiple adjacent loops on the needle.
Purl two together – p. 2 t., p2to, pto
Purl two together was a very common instruction, especially when columns of fagoting style lacy knitting were used.
Crossed knit – c, t, b, tw
Crossed knits are modern twisted knit stitches, produced by knitting into the back of a stitch (ktbl). I haven’t seen a historical pattern that includes a purl through the back of the loop (ptbl), but that doesn’t mean that one doesn’t exist.
Now with all this set out, I can graph up the diamond insertion from the Delineator. It starts out with a cast-on of 23 stitches. It includes double YOs, but all double YOs are followed by p2tog units, producing the columns of fagoting in either side of the center design. I’ll show the progression from as-described rows through modern notation.
First, as written, preserving the double YOs; without flipping the wrong-side rows in accordance with the modern charting convention of showing the work as it appears on the front (public) side; and without centering the rows or norming the chart to have parallel edges we get this: (click on images below for larger versions):
We quickly see that stitch counts vary from row to row, although the pattern is more or less internally proofed because wrong side rows do contain the same number of stitches as the right side rows that preceded them. We also see that the double YO followed by P2tog problem is here. Were those YOs to each be “real” each following wrong side row would need to be two stitches longer, and the lacy effect would not be achieved.
Other features of this pattern are pretty straightforward. YOs are YOs, whether they appear on the front or reverse side rows. The K3tog unit only shows up on front (even) side rows. P2tog when seen from the back is a plain old k2tog, so that’s also easy to flip.
So. Norming the presentation so that wrong-side rows are shown using the correct right-side row equivalent symbol, and isolating the side columns of fagot stitch, and consolidating the YOs we get:
I’ve gone through all of this not only for the fun of sharing, but also because I am using this particular pattern to knit up a new quick lace scarf. I’ll edge the thing out with something complementary, but for now, here’s how it looks:
Knit somewhat overscale in Swift River Prescott on US #8s, one panel of this lacy pattern is perfect for a scarf – curl-free and totally reversible!
Here’s the last item in the chart series. Thank you for all the kind
words. I’m delighted that people are finding this useful.
have gotten some questions about why I am not using the standard
Japanese symbol set. That set is quite broad compared to most of the
sets in Western books. My answer is that it’s relatively unknown in the
US and Europe. Perhaps I’ll add a symbol glossary that equates its
symbols to notations used by other more commonly available sources.
That’s a big project though, and might be better suited for wiseNeedle
than for this blog.
Barbara Walker’s Starlight Lace, Second Treasury of Knitting Patterns, p.288
will use this last Walker pattern to show some more complications to
charting life. This time, the pattern’s stitch count varies in a couple
of rows, plus there is a large number of edge stitches. My method is to
graph out everything verbatim row to row, then (if needed) introduce
no-stitch boxes for clarity. Again, all quotations from Walker are in
bold. Here goes…
Multiple of 6 st plus 5
Row 1 (wrong side): and all other wrong-side rows – Purl.
Row 2: K2 *yo, ssk, k1, yo, ssk, k1-b; rep from *; end yo, ssk, k1.
repeat is only 6 stitches, but I think I’ll chart out three repeats plus
edge stitches. That should give me enough room to see the play of the
edge stitches, and the staggered effect of the offset design itself.
Row 1 is plain old purl, but it’s a wrong side row, so it graphs out as
shown below, with the “1″ on the left hand edge rather than the right
hand edge. Remember, I’m just graphing verbatim at this point. I’m
making no effort to read ahead. I just want to get the stitches down on
got 23 stitches [(6x3) + 5]. Note that the k1-b (knit one stitch
through the back of the loop to twist it) has its own symbol. All wrong
side rows in this pattern are plain – worked as purls if the thing is
knit flat, and as knits if it’s worked in the round.
Most modern texture patterns alternate rows with something happening on
them (cable crossings, decreases, increases, etc.) with plain rows, and
many pattern authors don’t bother graphing the alternate rows if
they’re all plain. This can cause a bit of confusion. I got tripped up
recently by Hazel Carter’s Spider Queen shawl. It’s a masterful bit of
charting, but the first chart is stripped of those plain wrong-side
rows. The later charts include them. I wasn’t paying attention, and
didn’t notice that the numbering on that first chart labeled every row,
but counted by twos. I ended up having to rip back a bit when I noticed
that my piece didn’t looklike the project photo. So be warned. Look
at the numbering. If it begins with “1″ on the left, and you’re
knitting flat you start off with a wrong-side row. If the “1″ is on the
right and you’re knitting flat, you start off with a right-side row. If
every other number is missing, you’ve got a pattern with the plain rows
left out. Look elsewhere in the write-up to find out if those plain
rows are to be knit or purled.
I’ll skip writing up the plain rows, but I will include them in my growing graph:
Row 4: K3, *k2tog, yo, k1-b, yo, ssk, k1-b, rep from *; end k2
no problems here. Everything graphs out nicely and stitch count is
constant. There are equal numbers of stitches increased (the yos) and
stitches decreased (the ssks and k2togs).
Sometimes if I’m
having problems with a repeat, even if it’s charted, I’ll grab a piece
of graph paper and draw out my stitches. Sometimes I catch an error in
my knitting using my pencil that went totally unnoticed on my needles.
Row 6: k2, k2tog * yo, sl2-k1-p2sso, yo, sl1-k2tog-psso; rep from * end yo, sl2-k1-p2sso, yo ssk, k2.
is where that “off to hell in a handbasket” feeling begins to creep in.
We’ve got double decreases, both with the rightmost leg on top
(sl1-k2tog-psso), and with the centermost stitch on top (sl2-k1-p2sso).
We’ve also got a number of yarn overs, and just for fun – a couple of
plain old decreases, and an unknown number of times to do the ** repeat
between the k2, k2tog opening unit, and the end yo, sl2-k1-p2sso, yo,
ssk, k2 closing unit.
To figure this out, we need to remember
that we’ve got 23 base stitches on the previous row. That’s 23 stitches
to play with. All of the plain knits plus the stitches in the decreases
on Row 6 must add up to 23. Let’s look at the math:
- One ** repeat on this row adds up to six stitches (the two double decreases).
- The pre-** opening row unit is four stitches (k2 plus one k2tog)
- The after-** closing row unit is seven stitches (one double decrease plus one ssk and k2)
you add up our fixed numbers (the pre- and post-** stitches) you get 11
stitches. The previous row contained 23, and we subtract those 11 from
the total. We get 12, which (serendipity) is a multiple of our ** unit.
We graph out the pre-* stitches (shown in blue) plus two repeats of the
** unit, followed by the post-** unit (also shown in blue.
We’re out of that handbasket, even though our graph is showing a very
short row. Not to worry. Going through and counting stitches confirms
that we’ve got the correct number here. We’ll worry about neatening
everything up and inserting those no-stitch boxes after we get all the
rows charted. So let’s move on.
Row 8: K3, *k1-b, yo, k1, yo, k1-b, k1; rep from*, end k2.
row is also problematic. How many times to repeat the stuff between the
**s? Again , stitch count comes to our rescue. Evil Row 6 brought the
stitch count down to 17. Row 7 (worked plain) preserved that count. Now
on Row 8, there are increases, and “as-is” stitches but no decreases.
There should be 17 stitches on this row EXCLUSIVE of the YOs. Again we
do the math. We start with 17 stitches, then account for the three
before the *, and the 2 after – that’s 12 stitches left. NOT counting
YOs, each between the ** repeat contains 4 stitches. We need to graph
out three iterations of the stuff between the **s. Happily once we
graph in these instructions (including the 6 YOs) that restores us to
the original stitch count of 23.
we’re not worrying about lining stitches up right now, our only concern
is getting the correct number of them on the chart. We’ll think about
how to represent those low-count rows 6 and 7 later.
Row 10: K2, *yo, ssk, k1-b, yo, ssk,k1; rep from * end yo, ssk, k1
back to a stable stitch count, with the same number of increases and
decreases per row. Graphing it up is easy. I notice something here
those two blue units? They’re identical. It looks like this pattern is
formed by an exact duplicate of rows 1-6, offset by three stitches (one
half of the repeat). While you can see it (sort of) in the prose
directions, the duplication leaps out in the charted ones. I find this
sort of half-drop duplication and charting makes the pattern really
easy to memorize. More on this later, after we’ve charted some more
Row 12: *K2tog, yo, k1-b, yo, ssk, k1-b; rep from *, end k2 tog, yo, k1-b, yo, ssk.
Again this looks veeerrrryyyy familiar! I’ve highlighted the repeat (in fact I just cut and pasted those boxes).
Row 14: K1, *yo, sl2-k1-p2sso, yo, sl1-k2tog-psso; rep from *, end yo, sl2-k1-p2sso, yo, k1.
Evil Row 6, with all those double decreases? It’s back! Offset three
stitches, but otherwise the same. We start with 23 stitches on the
previous row, then subtract the 1 before the **, and the 4 after the
**, leaving 18 stitches – so we do the 6-stitch bit between the **s
Row 16: K1, k1-b, *k1, yo, k1-b, k1, k1-b, yo; rep from * end k1, k1-b, k1.
like row 8, offset again by three stitches. Again we’ve got 17 stitches
on the previous row to account for. Not counting the YOs, we’ve got 2
stitches before and 3 stitches after the ** accounted for, leaving 12 -
so we do the 4-stitch ** unit three times. One you add in the YOs,
we’re back up to to the 23 stitches of our original count.
to add the finishing touches. It looks like each of the decrease units
on Rows 6 and 14 visually caps off the clusters of decreases on the
rows below. So I’ll spread them out across the row, adding in my
no-stitch boxes as best I can to maximize the read of the pattern
compared to the photo of the worked swatch.
I’ll also add in
my stitch key, header and footer info at this point. Remember that
there are NO increases or decreases on alternate rows. Therefore I
don’t need to include that second column of “if it’s a
right-side/wrong-side row” instructions that I had to include in
One final note, there is one small bit of
strangeness here. Because of the way that the repeat works out, and the
way that edge stitches are handled, the last decrease on Row 6 is
handled differently if it is the final stitch of an “inside repeat” or
if it is the final stitch of the last repeat on the row. Since this
isn’t easy to graph, I’ve added a special note about it, and made it
blue on the chart.
The memorization thing?? This pattern looks complicated at first glance. Especially if you just look at the prose directions. However it’s not that tough. There are only four substantive rows – 2, 4, 6, and 8. The entire pattern repeat is only six stitches wide. Everything else is a repeat, either straight on the same row, or (in the case of rows 10-16) offset by three stitches (one half the width of the repeat). ? I can’t remember the prose directions verbatim, but I can and do memorize the pattern in its visual representation. Not everyone can memorize a nonverbal visual representation (and it’s no shame not to have that bit of wiring) but many people can, and have surprised themselves by being able to do so after becoming comfortable with charts.
concludes my mini-series on graphing – how to read them, how to build
them, and how to solve common problems translating prose directions to
charts. Please feel free to post additional questions about graphing
and reinterpreting prose instructions as charts, but please know I will
not be offering a graphing service here. My goal is to show others how
to do it for themselves, not do it for them.
One last tech note
- the visual presentation of the charts changes mid-way through this
note because I experienced a massive computer failure. I ended up
finishing this post on a different machine using a different version of
MS Visio. The later version has a slightly different GIF translator
than the earlier version I normally use. So it’s not your monitor -
it’s me. Apologies for the visual confusion.
So far the nominations for stitches to use as object lessons have been rather sparse. I’ve gotten suggestions to do:
- Porcupine Stitch from B. Walker’s Second Treasury of Knitting Patterns, p. 282
- Drooping Elm Leaves from B. Walker’s A Treasury of Knitting Patterns, p. 217
I’ve also gotten notes from people who said that given the hints posted over the past week they’ve been able to graph up
- Mermaid Mesh from Walker’s Second Treasury, p. 267
- Madeira Cascade from Walker’s A Treasury, p. 222
As the big boss at work would say, “Good on ‘ya!”
two patterns are not quite straightforward. Cascade has five stitches
above and beyond the repeat that need to be apportioned into edge
stitches. It does however have a very strong central spine – a double
decrease that lines up on all right-side rows. Mesh is a bit harder in
that it has both lots of edge stitches, plus a massive number of
decreases and increases that use natural slant of the decreases to
visually wander left and right. Certainly not a pattern for the
faint-hearted to graph!
For the object lesson I’ll do Porcupine and Walker’s Starlight Lace (Second Treasury,
p 288). Drooping Elm is interesting, but doesn’t pose some of the
conundrums that these two do. I’ll start today with Porcupine.
Starlight will appear later in the week.
Porcupine Stitch from B. Walker’s Second Treasury of Knitting Patterns, p. 282
has some interesting features. It’s a 9 row repeat, in which only three
rows are substantive. BUT those three rows are each repeated at least
twice, and the same instructions are repeated on both the right and
wrong side rows of the piece. There are also four stitches requested
over the 12 stitch repeat
count that will have to be accounted for in edge stitches, but they
seem to always stay outside the ** repeat marks, so keeping track of
them shouldn’t be a problem.
Walker notes that this texture design is of Victorian origin. It
does have a major feature that was much more common in early
instructions than in later ones. Porcupine includes patterning on both
right side and wrong
side rows. You don’t see this often as most modern? patterns
confine increases, decreases or other shaping elements to right-side
rows only. Flipping the instructions for decreases is far more
confusing than just translating knits to purls and vice versa.
write-up will intersperse the as Walker gives them with how that row
ends up being graphed. The Walker quotations will be in bold.
Multiple of 12 stitches plus 4
Row 1: K2, *Yo, K2tog; rep from * to last 2 sts, end k2
at the pattern, I suspect it will be a good idea to graph out two
repeats of the pattern, that’s 24+4 = 28 stitches across my chart. We
start with a right-side row:
Row 2 and 4: K2, purl to last 2 st, end k2
easy. Remember this is a wrong side row, and that mental inversion
thing should be invoked to “flip” knits to purls and vice versa.
Row 3: Knit
Because Row 4 is the same as Row 2, I’ll graph up both 3 and 4 here.
Row 5 and 8: K2, *sl1-k2tog-psso, k4, yo, k1, yo, k4, re from *, end k2
it begins to get interesting. Still, stitch counts are maintained. How
can I tell this? By looking at the part between the **s. It includes a
double decrease that finishes with the rightmost stitch on top, plus
two yos to compensate for the two stitches eaten by the double
decrease. Warning though. It’s not all that hard to visualize row 5,
it’s a right-side (odd numbered) row, but I can sense some
hyperventilation among those who have noticed that this same sequence
is repeated on a wrong-side (even numbered) row. We’ll deal with that
bit of chaos when we get there.
Row 6, 7 and 9: K2* p3tog, p4, yo, p1, yo, p4, rep from * end k2
have now hit the twilight zone row – the one that will cause many
people to give up graphing. But it’s not impossible. Remember that
mental flip thing? Flex your brain because we’re now going to do some
On Row 6, we’ve been told to do a p3tog on a wrong
side row. Now, a p3tog on a wrong side row, if viewed from the right
side of the work is a dead ringer for a k3tog. How do I know this? The
Sainted Barbara tells me so in the glossary of chart symbols in her Charted Knitting Designs (aka Walker III), and A Fourth Treasury of Knitting Patterns
(aka Walker IV). Also I experimented. I’ll use my symbol for k3tog, BUT
I’ll remember to build a double column glossary to accompany this
pattern that describes what should be done when this symbol is
encountered on both right-side and wrong-side rows.
on Row 7, we’re told to do the same thing as on Row 6. But we’re on a
right-side row. A p3tog on a right side row is a p3tog on a right side
row. I don’t have a symbol in my set for a p3tog, so I’ll have to make
one up. Visually, in a P3tog done on an odd numbered row, the right
hand most stitch of the three worked together ends up on top. I’ll make
a hybrid symbol that sort of reminds me that three stitches are being
worked together, the right hand most one will end up on top, and that
it’s a purl. If it turns out that I like this symbol, I’ll add it to my
permanent stencil collection in Visio:
Row 5 and 8: K2, *sl1-k2tog-psso, k4, yo, k1, yo, k4, re from *, end k2
8 is a repeat of Row 5, but it’s done on a wrong-side as opposed to
right-side row. Again referring to the Sainted Barbara, we see that a
s1-k2tog-psso done on the right side has as its wrong-side counterpart
the delightfully awkward p3tog through the back of the loop. Again -
remember we don’t actually have to DO a p3tog through the back of the
loop here unless we are doing this pattern in the round, but the symbol
we use on the chart is the same one that would be used for one of those
awkward puppies worked on the right side. I don’t happen to have a
standard symbol for p3tog through the back of the loop, so I’ll invent
Row 6, 7 and 9: K2* p3tog, p4, yo, p1, yo, p4, rep from * end k2
9 is a duplicate of Row 7. We’ve already graphed that. So we now have
the nine rows of our repeat. It’s also become clear that stitch counts are rock-stable row to row, and that the four
extra stitches here are just garter stitch selvedges there for
convenience, and aren’t required to eke out partial repeats of
the pattern. I’ll mark the four extras off in blue.
But we’re not quite done even though all nine rows are graphed out.
got a repeat made up of an odd number of rows. That means that Row 1
repeats on Row 10. In fact, although rows 10-18 are the same as Rows
1-9, each one graphs up as its opposite-side sibling. (I can sense I’ve
lost quite a few of you, so I’ll show rows 10-13:
10 duplicates the action of Row 1, but does it on a wrong-side row.
Therefore, the stitches that graph up as K2togs in Row 1 use a
different symbol in Row 11. Likewise the knits/purls of rows 11-13 show
as their opposite.
Row 14 duplicates Row 5, but as a
wrong-side row. We’ve already graphed that bit of twisted thinking on
Row 8, so adding it isn’t a problem. Row 15 replicates Row 6, again we
already did that flip on Row 9, so a simple cut and paste takes care if
16 duplicates Row 7, which has its wrong-side counterpart originally on
Row 6. Row 17 is another Row 8 in its right-side expression (Row 5).
Row 18 is another Row 9 flipped for the wrong side (Row 6). If you
place all of them on the chart, add the stitch key, grids, titles, and
attributions you end up with this:
this may seem a long way to go for a short drink of water compared to
Walker’s original write-up. In this case, the prose description is only
five lines long, but the chart takes up half a page. There’s no bonus
for brevity awarded for the charted format. But there is one
major advantage to having this described in a graph. This
chart is equally useful to people knitting in the flat and people
knitting in the round, because all the right/wrong side transformations
People knitting in the round experience every row as a right-side
row. To knit this reversible pattern entirely in the round,
they’d cast on an even number of the stitch multiple (without the four
blue extra stitches) then they’d follow every row starting at the right
hand edge of the graph, and using the key symbols as interpreted
in the “On Right-Side Rows” column. People knitting in the
flat would follow the chart in the manner I described before, starting
the odd numbered rows at the right edge, and the even numbered rows at
the left, alternately using the appropriate columns from the
accompanying symbol key.
Have fun with this one. Try out Porcupine Stitch in a swatch.
You’ll find the lacy effect is magnified if a larger needle than one
would usually use for a given yarn is used. Lacy or dense, the
result will be rather puffy. Given the appropriate yarn it would
make nice two-sided scarves, shawls, or blankets. Stay tuned for more adventures in charting!
In a perfect world there would be an intuitive set of graphing symbols
that would be quick and easy to understand. They’d cover all possible
maneuvers in knitting, and would be useful in every circumstance. This
is however, a total pipe dream. Knitting is near infinite, and
knitters are fiendishly clever in the variant ways they have found to
produce their desired results. There are a number of knitting
techniques and stitches that pose special problems to charting:
Large numbers of stitches increased or decreased at the same time
When you see instructions like “make 5 in next stitch” you’ll need to invent
a symbol to handle it. I’ve seen German and Japanese charts that use a
variant on something like this:
Decreasing a large number into one stitch would generate need for something
similar, perhaps with the V upside down, and the number of stitches to
be eaten indicated between its open toes.
Also unless you’re dealing with an edging, it will probably be impossible to graph up a
pattern containing massive group increases or decreases without using the
no-stitch boxes we discussed yesterday. Still, these problems fall into
the “inconvenient but not insurmountable” camp.
Bobble and bell-shaped semi-detached units
Some bobble and bell units are produced by knitting back and forth over a
small number of stitches, to make a blister-like addition that’s
attached to the main work at top and bottom. Most chart authors treat
this type of unit as a separate sub-process. The main chart may have a
single box with a specified symbol in it, indicating where the unit is
to be placed. The unit itself will be described either in prose, or in
a “mini-chart” accompanying the main chart as a sidebar. Another
“inconvenient but not fatal” challenge.
Patterns containing stitches either slipped from or knit into the row below
These can pose real charting problems, especially in linen stitch family
textures where large numbers of stitches are worked “out of row.” I’ve
seen large V-shapes superimposed on the graph that are supposed to
represent these distended stitches, but they are visually difficult to
deal with. If there are lots of them, the clutter can be overwhelming,
and some linen stitch or slip-stitch based patterns may be impossible
to graph at all.
It is interesting to note that B. Walker used a special charting notation for her slip-stitch based mosaic colorwork. In that format each row of the chart represented two rows of knitting
instead of the more conventional one row worked = one row charted
ratio. She didn’t try to show stitch deformation by the use of a symbol
set, instead she stuck to two-color mosaic patterns that swapped colors
every two rows. The squares on her charts indicate whether one is to
form the next stitch by working with the current strand, or slipping
the color of the previous two-row set up onto the needle.
Threaded stitches or stitches with right-side floats, or decorative wraps spanning one or more stitches
There are some patterns that form colorwork or texture patterning by using
separate strands that are threaded back and forth through live stitches
during knitting. Other patterns use as decorative elements floats or
wraps of one or more stitches, deliberately formed on the right side of
the work. These are both very difficult to represent in charts. I’d
probably go with some sort of notation in the main chart that Effect #1
happens here, and accompany the chart with a separate detail write-up.
Some popular novelty stitches are near impossible to chart. Loop Stitch is a
good example. That’s the stitch used to make a surface completely
covered in shag-rug style loops. The manipulations required to make the
loops don’t lend themselves to graphing, and beyond noting which
stitches carry the loops in a piece that uses both adorned and
unadorned areas for contrast, indicating their presence is of little
In spite of these exceptions, if a pattern contains just knits, purls, cables, simple increases and decreases – even twisted stitches – it can probably be graphed. The graph may be massive, but it can be done.
I’ve got only one nomination for a particularly vexing pattern to use in tomorrow’s object lesson. If you’re got one to suggest, please send me an eMail (replace the “AT” in the address with the standard @ sign).
We’ve covered basic charting, and charting variable width edgings. Now
for panels and insertions. Those are patterns that can be used as
accents in the main body of your piece. Sometimes they show up as
single strip scarves, sometimes several repeats of the design are
combined across to make an all-over design (occasionally fitted
together with half-drop variations), sometimes a single panel is
repeated to make a long stripe in combo with a stockinette ground,
sometimes just one vertical repeat of the design is used as a spot
accent, sometimes panels of different patterns show up side by side.
What makes them different from edging patterns is that they can be
embedded in the center of a piece, and that piece can be knit in the
These insertion style patterns can have either stable or variable stitch
counts from row to row. One with a stable count (either no
increases/decreases or an equal number of increases to decreases on
every row where they occur) are graphed more or less the same way as
the pattern in Charting 101. The ones with changing stitch counts do
pose special problems.
Let’s consider this simple variable count
insertion. It’s my own write up of a simple embossed leaf inside a
framing K2, P2 rib:
Cast on 9
Row 1 (wrong side): P2, K5, P2
Row 2: K2, P2, (K,P,K in one stitch), P1, K2
Row 3: P2, K2, P3, K2, P2
Row 4: K2, P2, (K1, YO)2x, K1, P2, K2
Row 5: P2, K2, P5, K2, P2
Row 6: K2, P2, K2, YO, K1, YO, K2, P2, K2
Row 7: P2, K2, P7, K2, P2
Row 8: K2, P2, K3, YO, K1, YO, K3, P2, K2
Row 9: P2, K2, P9, K2, P2
Row 10: K2, P2, SSK, K5, K2tog, P2, K2
Row 11: P2, K2, P7, K2, P2
Row 12: K2, P2, SSK, K3, K2tog, P2, K2
Row 13: P2, K2, P5, K2, P2
Row 14: K2, P2, SSK, K1, K2tog, P2, K2
Row 15: P2, K2, P3, K2, P2
Row 16: K2, P2, K3tog, P2, K2
As you can see, the thing starts out being nine stitches across, but grows on row 9 to 17 stitches across.
How to chart? The symbol set is pretty straightforward. Each
individual row poses no problems. For example, here’s row 8:
If we normed one edge like we did with the edging patterns, we’d end up with this:
While all the info is there and this chart could be worked from, it’s
deceptive in that it looks like an edging. Plus one of charting’s prime
directives – representing knitting in a format that’s visually akin to
the finished product – has been fouled.
So. Let’s look closer at this pattern, looking for obvious points of internal symmetry or
reference. We quickly see that the thing IS symmetrical. There’s a
center stitch in every row. Let’s stack our rows on the center stitch:
That’s closer. You can begin to see the leaf shape in the center,
but the wiggly edges are still a bit confusing. Here’s another
cut at the same basic concept. This time however, I’ve lined up
not only the center stitch, but also the knit ribs that frame it:
Those gray areas? They don’t exist. Flat out aren’t
there. They’re the equivalent of the stage attendants dressed in
all black who move props around in full view of the audience during a
drama or puppet performance. You’re not supposed to see them, even
though they’re in plain sight.
The grayed out areas are spacing mechanisms introduced for the sake of
visual clarity in the rest of the pattern. They have no
correlation to stitches in the actual knitted piece. Working from
this chart, I’d skip right over the gray background. My first row would
be P2, K5, P2, just as in the written directions. Now different
authors represent non-charted “no stitch” or null spaces
differently. I chose to use a general background shading, with no
boxes marking individual stitches. Other people don’t bother
removing the box notation from the no-stitch spaces. On their
charts the no-stitch boxes can be a bit harder to interpret.
How to know when to use mystery no-stitch boxes? Although it’s a
matter of personal preference, sometimes they’re absolutely necessary
because there just isn’t room to graph out your piece unless they’re in
the mix. I could graph out my embossed leaf without the
no-stitch areas, but if this leaf was part of a larger graph covering a
wider area, the distortion introduced by the width of the longest row
might ripple out and perturb the representation of design elements to
either side. In that case, using the no-stitch boxes would keep
my two edges parallel and let the leaf panel sit more comfortably in
the total project chart. That in turn would help the
knitter keep his or her place on the wider graph.
Tomorrow I’ll look at patterns that are extremely hard (if not
impossible) to chart out. The final piece in this series I’ll
build one chart for a lacy or complex cabled design that has presented
a special challenge. Nominations for the final object lesson will
be accepted. Please contact me off-list before Thursday night if you know of a prose texture pattern you’d like to suggest for group edification.
I’m delighted that people found yesterday’s post useful. The most asked
question though was TexAnne’s original one – what does one do when
stitch counts change from row to row?
The guiding principle here is clarity of illustration. You want your chart
to reflect as closely as possible the visual appearance of the finished
knitting. That might mean that you handle the problem differently
depending on the general situation. An edging for example might be
graphed up differently than a panel insertion.
Let’s start with the basics – some different types of increases and decreases. They are
after all the Evil Agents that perturb stitch count across rows.
Increases come in two flavors – visible and invisible. A visible increase is
something like a yarn over. It’s an increase that leaves an intentional
eyelet hole in the piece. Invisible increases come in all sorts of
flavors – some more invisible than others. Often an invisible increase
is called a make one. Some people favor raised bar style increases,
others do the knit into a stitch of the row below, and others go for
the less invisible knit/purl into the same stitch (or k1 front, k1 back
into the same stitch). Which method of invisible increase is used is up
to the knitter, although the designer may suggest one that works
particularly well for the project in hand. In general though, the two
types of increase have different notations in charting. I favor a boxed
circle for a visible increase and a boxed M for an invisible increase.
I even go so far as to slap a bar across the M if I want to
specifically call for an invisible increase that forms a purl stitch.
Here are the symbols I use for some of the more common increases and
Apologies for the size of the illustrations today. I’m having an argument
with the picture upload facility, and this is the best resolution that
I can get working this morning. Although my symbols were inspired
by B. Walker’s and L. Stanfeld’s two, apparently I stuck to industry
standard practice, doing whatever the heck I felt like doing and coming
up with my own set.
Now. How do you use these?
Let’s start with a simple edging. Edgings generally have one straight edge
where they attach to the thing being trimmed, and one that’s dagged,
pointed, crenelated, scalloped, picoted or otherwise fancified in some
way. The fancy bits (I’ll call them all points for simplicity of
reference), are formed by increases and decreases. In some the
decreases come as partially bound off rows. Here’s a good example. This
one is the edging I used on my Kombu scarf:
Cast on 4
Row 1, 3, 5 (wrong side) Knit
Row 2: K1, YO, K1, YO, K2
Row 4: K2, YO, K2tog, YO, K2
Row 6: K3, YO, K2tog, YO, K2
Row 7: Bind off 4, K3 (4 stitches remain)
Repeat rows 2-7
First off – assume that all edgings knit side to side are knit in the flat.
The ‘wrong side’ notation confirms this. Row 7 starts with binding off
and is a wrong side row. That means that if you hold the piece with the
RIGHT side facing you, the straight edge will be on your right, and the
ziggy-zaggy one will be on your left. Row 1 is not repeated, and
appears to be just a foundation row. Armed with this orientation info,
we begin charting rows 1 and 2.
Row one is a wrong-side row, so even though the directions say “knit” the
stitches are plotted as purls (that chart = right side view
thing). On row two we’ve got two yarn overs. They increase
the total stitch count by two. We know that this is an
edging. We know that the jagged edge will be on the left when
viewed from the right side. Therefore I have chosen to make the
stitches line up along the right hand edge. Here’s a proofing
trick. There are no decreases in this row, therefore number of
stitches in this row EXCLUSIVE of the yarn overs should be equal to the
number of stitches in the row below. 4=4, we’re o.k. Let’s
Row 4 contains two yarn overs, but the total stitch count is
increased by only one block. That’s because it also contains a
K2tog (stitch count +2yos -1k2tog = stitch count +1, not +2.) Row
5 is just another all knit row. Proofing stitch counts vis
a vis the previous row can be done by counting the number of plain
knits, plus two for every K2tog or SSK decrease. In this case
we’ve got 6=6. It works.
Row 6 and 7 get interesting. There is no uniformly acknowledged
(or obvious) symbol for binding off, therefore charts that contain
bound off stitches often use a text notation to indicate what’s going
on. Also remember if you bind off stitches you end up with one
remaining loop on the needle:
Again Row 6 increases total stitch count by only one (two steps
forward, one step back covers the YOs and the decrease). Row 7
includes the instruction to bind off four stitches, BUT there’s a
visual discrepancy between the chart and the written directions.
The chart says BO4, K3. The chart shows four stitches.
That’s because one of those is the loop that remains after you’ve bound
off the four stitches at the beginning of Row 7. You have that
loop on the needle, then you knit the remaining three stitches, for a
total of four stitches on the needle. I’ve also shaded out Row
#1. Just like the edge stitches in yesterday’s illustration, this
indicates that Row 1 is not part of the regular repeat. It’s a
foundation row worked at the start of the edging strip, and not
As you can see, simple edgings are relatively easy bits of lace to
graph. Stitch counts do vary from row to row, but because they
have a stable edge, those extra stitches have someplace to go,
visually. Having built this foundation of basic concepts,
tomorrow we’ll do a panel pattern that doesn’t have the luxury of
a free edge, and introduce The Stitch That Isn’t There bugbear.
TexAnne eMailed a question that sends me off on a tangent. She’d like
to know more about how to take a set of prose directions from a source
like one of Barbara Walker’s stitch treasuries and turn them into a
chart. In specific, she’d like to know about how to handle things
like double yarn overs, and stitch counts that vary from row to
row. These are excellent questions. Since not everyone can
leap right in at the graphing lace level, I’ll start with simple
charting and work up to the harder bits later in the week.
start, transforming prose to charted directions is easier than some
people think. Tools include some sort of mechanism on which to do the
chart. I ping-pong back and forth between good old pencil and paper,
and Microsoft Visio. Visio is an expensive Windows drafting/drawing
program that I have on hand, mostly because I use it when I work as a
consultant. There are other solutions out there, ranging from forcing
spreadsheets to handle the function, to dedicated knitting programs.
But don’t despair if you have access to no computerized tool for
charting. Plain old 1/4 inch quadrille paper (the Junior High
School geometry teacher’s friend), a friendly pencil and forgiving
eraser work just fine. Principles of conversion remain the same
regardless of tool used.
Let’s start out with some basics.
Charts are read from the bottom up. In most cases (but by no means
always) there is a one to one correspondence between a stitch in the
work and a box on the graph.
Charts represent the work as seen
from the public or right side. As such, if you’re working flat, you
need to remember that the same symbol that represented a knit stitch on
your right side row will represent a purl stitch on your wrong side
row. (If you’re unsure of this basic binary truth, go grab something
with both knits and purls on the same row, like a swatch in ribbing,
and use a pin to poke through a single stitch, then identify it on both
sides of the work).
Two more basic truths of charts:
- Almost every author or chart source has a unique symbol set. Some are similar, but none are absolutely identical.
- Not everything can be completely charted
Oh, chart purists will argue about #2, but there ARE some patterns that just don’t lend themselves well to charted expression.
start with an easy one. Here’s a recipe for a simple basket rib done in
all knits and purls. Stitch counts remain constant from row to
row. The source is Barbara Walker’s Treasury of Knitting Patterns, Scribners, 1968;? page 17 (see footnote below).
Multiple of 4 stitches + 1
Row 1: (right side) K1 , *p1, k1; rep from *
Row 2: K2, *p1, k3; rep from *, end p1, k2.
Row 3: P2, *k1, p3; rep from *, end k1, p2
Row 4: P1, *k1, p1: rep from *
Row 5: K1, *p3, k1; rep from *
Row 6: P1, *k3, p1; rep from *
to chart this out, we examine the instructions. It’s pretty clear that
there will be an edge stitch. The “Multiple of 4 stitches + 1″ says so.
So let’s start with Row 1. The stitches will read exactly as written,
in the direction of the work. That means that the first stitch will be
at the right hand edge of our chart. Since the directions call for a
multiple of 4, +1, let’s start off with an auspicious 13 stitches -
that’s three repeats. plus that one spare:
well and good. In my twisted logic, a blank square is a knit, a square
with a dash in it is a purl. Not everyone uses this notation. Some
people use a square with a vertical line in it to represent a knit, and
a horizontal to show a purl. Some people use a dot to indicate a purl.
There’s never been any international standardization of knit symbols,
so use what’s comfortable to you.
If you follow the charts that I’ve put up here and on wiseNeedle,
you’ll notice that I like to keep tabs of how many stitches are across
my row by using a red rule every five stitches. When I chart out a big
pattern, I set up a large red grid first and then populate it, but here
I’ll add in the red lines and row numbers as I need them. Again, this
is a matter of personal preference. Set your rules 4, 5, or 10 stitches
apart, or don’t use any at all. It’s up to you.
Now to add Row
#2. The original prose instructions were written for someone knitting
in the flat. In general unless you have absolute evidence to the
contrary from any accompanying text, assume that prose instructions are
written in the flat. This means that WHEN SEEN FROM THE FRONT OF THE
WORK, the second row will commence at the left hand edge of the graph.
A clue on this pattern is the notation “(Right side)”. In a piece knit
in the round EVERY row is a right side row, so this piece must have a
wrong side row – hence it is knit in the flat. Yes I know this is
confusing, because you always work in the same direction, but remember
that if you were knitting in the flat, you’d have flipped the work over
to go back.
A second complication! The prose instructions start off with K2, p1,
but the chart shows p2, k1! Don’t panic. Remember, we’re on the second
row – a wrong side (aka purl side or inside) row. The “2″ is at the
left edge to remind us of that fact. Those first two wrong-side knit
stitches WHEN SEEN FROM THE FRONT are purls. That’s the way they are
graphed. If your head is starting to hurt, just contemplate that while
this is a mind-stretching exercise, mental gymnastics like this have
been shown to delay brain aging.
Adding Row #3 makes which stitches compose the 4-stitch repeat more clear:
adding Row #4, I’ve moved to a more conventional method of shading.
Most charts that show edge stitches do so by shading them. Here it’s
clear that there are three repeats, plus one column of edge stitches
(to be fair, I could designate either the first or last column as my
edge, as in this simple pattern with a one-stitch edge, it doesn’t
matter which column serves that purpose.)
prose it’s not immediately evident where the actual repeat falls, and
what parts of the directions cover the non-repeating edge stitches.
This is one reason why I prefer working from charts.
covered the basics, I’ll quickly add the last two rows. Graphed out,
not only do we see where the repeats are, we also see that a
basketweave pattern is formed by a half-drop. Rows 1-3 and 4-6 contain
the same basic unit, but in rows 4-6 it’s advanced by two stitches. I’ve
marked the same basic unit in yellow on rows 1-3, and in orange on rows
simple nature of this repeat and the symmetry that builds it into a
basket weave pattern are difficult to discern by just reading the prose
instructions, but in a chart, the logic stands out.
knitters working both in the round and in the flat can use this same
chart. People knitting in the flat would cast on a multiple of 4
stitches plus 1. Then they would start at the bottom right corner and
work across Row 1, then they’d flip the work over and start the next
row at the “2″ – taking care to do the mental flip; and so on. People
knitting in the round would cast on an even multiple of 4 stitches, and
starting AFTER the blue edge stitch, would work across Row 1 as many
times as needed – skipping the blue stitches, eventually returning to
the point where their round commences. They’d then start to work Row 2,
again working from the chart’s right hand edge and skipping the blue
edge stitch whenever it was encountered.
So you see -
translating a pattern into a chart isn’t that tough. This particular
texture is an easy one. It’s all knits and purls, with no increases or
decreases. There are no variant stitch counts. Every row has the same
number of stitches. There are no slipped or dropped stitches, no wraps
or other bits of oddness.
In Charting 102, we’ll look at the
mystery that is The Stitch That Isn’t There. I’ll go over patterns with
increases and decreases, and what happens when the stitch count
changes. TexAnne, I hope this helps.
* My quotation of B. Walker’s directions, verbatim. Normally I
don’t do this. If I use a pattern that’s in a stitch treasury, I
try to alter it a bit. I start at a different place in the
repeat, center the repeat differently, chart it where it was in prose
before, or rechart it starting at a different point. I do this
because while no one person owns copyright on a knitting texture
pattern, they do own copyright on the way they have expressed that
pattern. This is analogous to recipes. No one owns the
concept of “apple pie,” but thousands of authors each own their
individual description of what goes into one, and how to make
In this case however, quoting Walker verbatim falls within the bounds
of fair use. I’ve given the citation, crediting the original
author. The quotation is there because the premise of this piece
is how to take a standard set of well-known prose instructions and turn
it into a chart.