It is summer again and I am sewing a lot of fine jersey. If you look through my blog, you’ll notice that I am rather obsessed with flexible seams – seams that stretch with the jersey instead of ripping under strain. I found that different machines, different mechanism types have varied success at this task, and I also found that it depends on the type of jersey you are sewing – its density, thickness and stretch, and whether it contains super elastic yarns like lycra, and in which proportion.
In this post I summarise what I found so far, and which adjustments on the machines contribute to getting that flexible stitch.
Types of jersey
There is any number of types of jersey, depending on how you look at it. I found the important characteristics to be the thickness of the material and the amount of lycra in it. Thicker materials are easier to sew – there is more bulk in which to coil the thread that can uncoil when the material is stretched, without putting strain on the stitching. So for example T-shirt weight interlock jersey is easier to sew than single knit jersey because of its thickness which is not always translated into weight. Fine quality interlock knit is light yet thick which feels very soft to the touch – it is made with very finely spun yarns. In contrast, ordinary quality single knit jersey is made with coarser yarns so it gives the same weight with lesser thickness because of the type of knit.
When lycra is added into the mix, it increases the elasticity of the material quite significantly, and not only across the width but also along the length. Knitted material made of cotton, viscose or bamboo fibres in itself tends to stretch a lot more across the width than along the length, which is known as a 2-way stretch (or sometimes 1-way stretch, to get you confused). Adding lycra can increase the stretch along the length to more or less the same level as stretch across the width, which is known as a 4-way stretch (or sometimes 2-way stretch – are you confused yet?). Depending on how much lycra is added, stretch levels can vary, so that the stretch along the length and across the width is not necessarily the same. For example, single knit cotton jersey with 5% lycra that I’ve used, had half the stretch along the length compared to across the width, while similarly made cotton jersey without lycra had only a fifth of the stretch along the length compared to across the width.
The type of fibre used to make the knitted material also contributes to the distribution of stretch. Vegetable fibres like cotton, viscose or bamboo give much more stretch across the width than along the length, whereas wool is the opposite – there is generally more stretch along the length than across the width. Synthetic materials like polyester or acrylic can stretch in both directions, while silk tends to stretch very little in either direction. What is also extremely important is how those fibres are spun – this can break all the rules. And of course mixing different fibres together in one yarn that is used to create that knitted fabric, changes everything again, so there is no fool-proof simple rule here. It is best to just test the stretch of the final product – the knitted material.
From experience with sewing knitted materials I found that the hardest fabric to get a flexible seam on is thin single-knit viscose jersey. These fabrics vary in their thickness, lycra content and stretch of course, it all depends on what you come across and what is available to you locally. The thinnest viscose jersey that I’ve used was less than a half of the thickness of the common T-shirt material, which also varies, of course! But I hope it gives you an idea. This thin viscose jersey typically has 110% stretch across the width and 40% stretch along the length, or if it has 2%-3% added lycra, it increases stretch along the length to about 60% without significantly increasing stretch across the width. I’m sure there are other thin and stretchy fabrics out there, but this is what turned up on my sewing table, and I use it as a benchmark for the ability of sewing machines to make a flexible stitch.
I found this thinnest jersey to be the worst case scenario for any machine, and they all could produce more stretch in the seams on thicker material. For example, even those machines that could not get any stretch at all on thin jersey, were often able to produce 25% stretch on a thick knit, with proper adjustments.
I am only looking at common lockstitch machines here, and although their mechanisms have different design, the machines I’m looking at all have the following components:
- One needle carrying the upper thread.
- One bobbin carrying the lower thread.
- A bobbin case or shuttle of some sort with thread tension adjustment for the lower thread.
- Upper thread tension adjuster.
- Upper thread take-up lever.
- A presser foot with a pressure adjuster.
- Lower feed mechanism: feed dogs under the cloth being sewn.
- Stitch length adjuster.
There are other types of lockstitch machines with different components. For example, early sewing machines didn’t have a take-up lever, or even didn’t have any feed at all! Some didn’t have any tension adjustment for the lower thread, while others had upper feed, notably Davis’ Vertical Feed. Some modern machines have more complex feed mechanisms, such as a walking foot (upper and lower feed) or differential feed. There are many other types of machines worth considering too, other than lockstitch machines, but all of this is for another post!
So, given the components in the list above, here are the adjustments that could be made to them. Not all of these adjustments are available in each machine!
- Upper thread tension adjustment.
- Lower thread tension adjustment.
- Foot pressure adjustment.
- Feed dog height adjustment.
- Stitch length adjustment.
- Take-up lever range adjustment.
- Take-up strength adjustment.
What makes for a flexible stitch
A flexible stitch stretches with the jersey because thread is being coiled in the thickness of the fabric during sewing. When the material is stretched, the thread becomes uncoiled so that the seam does not come under strain. Once all the coiled reserves have been straightened out, the stitches come under strain and the seam gives no more stretch. If you keep stretching it, the thread will break.
Obviously, the more coils you can fit per inch of cloth, the more thread is getting stored, so shorter stitches give more stretch in the seam.
Stitch length is either measured in the number of stitches per inch (SPI) or in millimetres (mm). Obviously, one is the inverse of the other: stitch length in mm equals 25.4mm (1 inch) divided by SPI. So for example if you have 21 SPI, this translates into stitch length of 1.2mm. Note that the stitches will appear shorter because some space is taken up by the point where the threads lock. The size of this point depends on the thickness of the thread, so you cannot make extremely short stitches with thick thread – it will be all points with no room for the stitches themselves!
Many machines have numbers shown near their stitch length regulator, and computerised machines even let you set stitch length on screen. Again, this can be either stitch length in mm or SPI. But the numbers will only correspond to the actual stitch lengths if the feed dog height is set just right, and this is not a fixed feature but an adjustable parameter. So if you want to use the numbers, check that they don’t lie first.
For sewing jersey I use stitch lengths between 21 SPI (1.2mm) and 40 SPI (0.6mm) with needle size 70 and plain household polyester thread (supposedly 120s but it’s so fluffy you can’t tell for sure). I could use better thread of course, and sometimes I do, but I have a drawer full of this plain stuff, and since this is the hardest thing to get a nice looking stitch with, that’s what I use for testing – we don’t want the test to be too easy! 😉
It is also easy to figure out that at the moment of stitch creation both upper and lower threads should not be very taut, or it won’t be possible to coil them. This is usually achieved by reducing thread tension, and anyhow, thread tension adjustments are always important in creating a flexible stitch. But there is more.
Foot pressure and needle point
The foot presses down on the cloth and prevents it from being pulled into the needle hole when the needle is going down, or being pulled up from the surface of the needle plate when the needle is going up. If the upper and lower threads are to be locked exactly in the middle of the thickness of the cloth, then the cloth has to be in the right place, and not above or below it. The type of needle point can also make a big difference here in preventing drag, in particular with materials made of finely spun fibres.
The foot pressure has to be high enough to ensure that the cloth stays in place, yet it must be low enough to prevent stretching out of the material during transport. It has to be relative to the height of the feed dogs because the feed dogs are pushing up the material while the foot is pressing down on it. If the foot pressure is too high and the material is elastic, then the feed dogs drag the material against the foot and stretch out that one bit instead of transporting the whole piece.
The exact mechanics of stitch creation differ between sewing machine designs, even though the basic principle of a lockstitch machine remains the same.
- The take-up lever pulls a length of upper thread off the spool.
- The needle penetrates the cloth carrying a loop of the upper thread in its eye.
- The hook under the cloth catches the loop from the needle and pulls it around the bobbin case or shuttle containing the bobbin with lower thread, thus intertwining (or locking) the upper and lower thread.
- The needle moves upwards and out of the cloth, pulling up the lower thread. This happens at the same time as step 1 – the take-up lever pulls on the thread too, simultaneously pulling a new length of upper thread off the spool in preparation for the next stitch, and pulling up the lower thread from the previous stitch.
Here is how Singer described this process on the example of Singer 15 sewing machine – but it is the same for other lockstitch machines as well.
Singer 15 illustrated above is a machine with a stationary bobbin case and a moving shuttle that wraps the thread loop around it. The bobbin case contains a round bobbin that spins when the thread is pulled by the take-up lever. In earlier variants of a lockstitch mechanism however, such as for example transverse shuttle or vibrating shuttle, the shuttle itself contains a long bobbin, and the bobbin spins because the shuttle moves away from the point of sewing.
The shuttle passes through the loop, pulling the bobbin thread through the loop with it, ready to be pulled up through the cloth by the take-up lever when the needle is going up. Notice the bobbin thread lying loosely on the shuttle at this moment, having been pulled off the bobbin by the movement of the shuttle arm.
The shuttle in these machines acts like a take-up lever for the lower thread – it pulls a length of thread off the bobbin in preparation for the next stitch. In machines with a stationary bobbin the lower thread is pulled solely by the upper thread when the needle is going up, there is no preparation for the next stitch. In my experience, long bobbin machines can deliver much better coiling of thread within the thickness of the material, and I believe it is exactly because the shuttle prepares a length of thread for the stitch similar to the take-up lever, so that the tension of the lower thread at that moment in time can be kept very low.
Take-up lever design with adjustments
The take-up lever pulls a length of the upper thread off the spool at the same time as it pulls up the lower thread after the loop has gone around the shuttle. Different designs of take-up lever make for different movement.
Transverse shuttle machines based on Singer 12 have a take-up lever that is pushed down by the needle bar and brought back up by a spring.
Inside the needle bar there is a rod that pushes on the take-up lever. How far the rod descends into the needle bar can be adjusted, and this determines how far down the take-up lever travels. As the take-up lever always rises to the same level pulled up by the spring, this rod in fact adjusts the range of motion of the take-up lever – the lower is the rod, the larger is the range of motion. This determines how much the upper thread is pulled up after the stitch is formed, and by consequence how much the lower thread is pulled up by the upper thread. The larger the range of motion, the more the threads are pulled up.
The spring at the back pulls the take-up lever back up when the needle bar rises. The strength of that spring determines the force of the jerk with which the lever jumps back. This jerk is mostly what pulls a length of thread off the spool ready for the next stitch – so the stronger the jerk, the more thread is prepared, and the lower is the upper thread tension at the moment of stitch formation.
On transverse shuttle machines the motion of the lever is synchronised with the needle: they reach the lowest point simultaneously. This allows to have that big loop of upper thread underneath the needle plate for the shuttle to pass through. This is shown in Fig. 2 in the description of Singer 15 above. The timing is different because it is a different mechanism!
The force of the spring and the range of motion of the take-up lever cannot be taken completely separately though – they act at the same time and they influence each other, both contributing to upper thread tension.
The manual for Jones Family TS machine stipulates that the take-up range adjuster should start pressing on the lever when the gauge mark on the needle bar is 3mm above the head of the machine. This is the same gauge mark that is used to set the needle. This setting results in a large take-up lever range, close to maximum. In my experience, a high take-up range works well for somewhat heavier or denser materials – from cotton poplin and up. The same manual recommends to increase the take-up range slightly for extremely thick materials – and a slight increase is all you can do until you hit the absolute maximum there. The manual does not say anything about reducing the take-up range below the recommended level. Well, I found that when you are sewing extremely fine materials like that thin jersey that I’m on about, a strong take-up range results in a seam that is too tight and not flexible enough. I reduced the take-up range quite a lot, to about half of the recommended range, and I also reduced take-up strength to the “light” setting (Jones Family TS has only two settings, some other machines have three). Reduced take-up means that with the same tension you get loops on the underside of the sewn material – the upper thread is not being pulled up quite as much. So I increased upper tension a little to compensate.
A word of caution however. Reducing take-up lever range can lead to skipped stitches, so make sure you don’t reduce it too far. Increasing foot pressure slightly often cures skipped stitches, as well as twisting the needle a little. Different threads and different types of needles require a twist either to the right or to the left, so experiment with yours. However, nothing will help if the take-up range is too low, there seems to be the lowest workable setting for a particular combination of fabric, needle and thread.
The best setting on Jones Family TS gives me 50% stretch at 21 SPI and 100% stretch at 36 SPI. This is without stretching out the jersey at all, with a good grip of those double-sided feed dogs.
On Singer 48K there’s less room for adjustment of the take-up range, but still I get 40% stretch at 21 SPI and 70% stretch at 36 SPI, all without stretching out the jersey and with neat stitching throughout. However, this is a one-sided feed dog, so you cannot sew bits with 4 layers of fabric to the right of the needle and a single layer to the left of the needle as is the case when sewing on a collar, for example. You are stitching through 3 layers but the fold is very narrow and the left-sided feed dogs only get to grip that single layer. This just doesn’t work and you have to turn your sewing around so as to have the thick bit on the left and the rest of the garment on the right and under the arm. Not ideal. Which is why I prefer Jones TS now which has double-sided feed dogs with equally good grip on either side of the needle.
Take-up lever design with fixed motion
Newer sewing machine designs have the take-up lever driven by a groove (cam) in the same disk that drives the needle bar, so the motion of the take-up lever is not only fully synchronised, but also completely fixed. There is nothing to adjust, it is always the same motion.
This is a diagram of the take-up mechanism of Singer 27 showing the cam determining the motion of the take-up lever. There is also a small take-up spring mounted on the thread tension unit. The strength of this spring can be adjusted, and it regulates the strength of the jerk with which a length of upper thread is pulled off the spool. But it is only a very small and light spring, so the adjustment doesn’t have much of an effect. From what I have seen with the adjustable take-up lever design described above, take-up range of motion has a much larger effect on stitch formation than the spring strength, and here in this new design the range of motion is fixed in the cam.
However, if you wanted to adjust it, note that what this spring does is two-fold: it reduces the jerk with which the thread is pulled off the spool in preparation for the next stitch while at the same time increasing the pull on the bobbin thread through the material that closes the previous stitch. The diagram above shows the “thread take-up spring regulation plate” which allows you to have this spring more or less wound, thus regulating its strength.
Setting this spring to strongest generally but not always gives more stretch in the stitching. When you increase the strength of the spring, you often have to reduce upper thread tension, and visa versa. Not all machines behave like that though, so check yours to be sure.
On vibrating shuttle machines the motion of the lever is timed very similar to the transverse shuttle machines but ever so slightly later – the lever reaches its lowest point when the needle is just starting to go up. In addition, when the needle is nearing its lowest position on the way down, the take-up lever pauses half way, so that the loop doesn’t get too big too early. It then quickly drops to the lowest point once the shuttle has caught the loop. A separate cam driving the take-up lever allows for this improved timing.
This new complex motion sounds very impressive, I must admit. All sorts of benefits come to mind. But the truth of the matter is that there was absolutely nothing wrong with the synchronised movement on TS machines, so what exactly are we improving here? There was no problem there to begin with! This improvement becomes, in my opinion, purely academic.
But what about the different makes of vibrating shuttle machines in my collection? Do they all have the same range of the take-up lever?
- Jones Medium CS: 4.6cm, no take-up spring.
- Clemens Müller’s Veritas VS2: 4cm, take-up spring.
- Bernhard Stoewer’s Serata VS2: 4cm, take-up spring.
- Bradbury Medium VS (VS2): 4.2cm, take-up spring.
- Mundlos 100 VS: 4.6cm, take-up spring.
Well, what do you know! The ranges differ, as do take-up springs – not only whether one is present or not, but how strong it is too. My Jones Medium CS has no take-up spring, and there are no provisions for it, so it does not appear to have been lost accidentally. Later models did get a take-up spring, as well as automatic tension release which my machine doesn’t have either.
This is the head of Mundlos 100 with its patented take-up lever drive. It is based on the same idea as Singer 27, but the motion pattern of the take-up lever is subtly different: it slows down significantly before reaching the top. The other machines raise the lever in one swing from the lowest point to the highest point, but Mundlos 100 raises the lever first from the lowest point to about 3.6cm mark quickly, then slows down near the top. So although the total range on it is higher than on Veritas or Serata, there is less of a jerk in pulling up the thread.
I have experimented with different settings of the take-up spring and different stitch lengths to see how it would influence the result.
This is the result from Veritas VS2. The left two stitches are done on the weakest spring setting, the right two stitches are done on the strongest setting. From left to right:
- Weakest spring setting:
- 21 SPI, 40% stretch. Good stitch.
- 40 SPI, 80% stretch. Stitching looks untidy and there are many skipped stitches. The jersey is stretched out. There are visible loop points on both sides. Not a great result.
- Strongest spring setting:
- 21 SPI, 60% stretch. Good stitch.
- 40 SPI, 90% stretch. Jersey is a bit stretched out and there are visible loop points on both sides. But there are no missing stitches and it is a passable result.
I did the same on Stoewer’s Serata VS2 which has the same take-up lever range but larger feed dogs, and consistent with my previous observations, I got a lot less stretch with the same settings. The strongest spring setting still gave the best stitch, but at 21 SPI I only got 40% stretch on the finest jersey. I say “only” but let’s not forget that most round bobbin machines get no stretch at all!
Setting the spring to a middle position gave intermediate results on both machines.
Overall, Serata’s style feels sharper somehow, more precise, and on woven fabrics Serata’s stitching is unsurpassed – she makes a truly superior straight and even stitch. Veritas feels softer and sloppier, and on woven fabrics she merely makes an excellent stitch, not supremely so. But she excels at stretch, as the tests show.
I haven’t got a Singer VS2 any longer, but I’ve had both models 27 and 127, and their sewing was somewhere in between.
Mundlos 100 showed a different behaviour, however. Both the strongest and the weakest spring setting gave less stretch than the middle setting. Also, contrary to Veritas and Serata, I did not have to rebalance tensions after changing the spring setting, and overall the spring had less of an effect here.
At 21 SPI I got 40% stretch on the strongest and weakest spring setting and 55% stretch on the middle setting. At 46 SPI I got 65% stretch on the strongest and weakest spring setting and 75% stretch on the middle setting. The quality of stitching was also quite poor on the strongest and weakest spring settings – it was quite uneven, with upper thread looking messy, but it was perfect at the middle spring setting, even at this extremely high SPI.
Mundlos 100 has rather large feed dogs but a very light foot pressure (it can be increased though). The feed is gentle and does not stretch out jersey too much – even at 46 SPI the stitching is still acceptable, and there is no stretching out at all at 32 SPI. This is better than both Veritas and Serata that start stretching out jersey at around 26 SPI on the lightest foot pressure that still makes good stitches. If the foot pressure drops too low, stitching becomes very uneven as the cloth is not kept at the right level during stitch formation.
Take-up lever design with reversed spring
This is in a way an intermediate design between the first two. Here the take-up lever rides on the disk that drives the needle bar, and in this it is similar to the Singer 27 design above. But the motion of the take-up lever is not fully determined by a cam, instead the lever is pulled down towards the disk by a spring mounted at its base, in the same place in fact as the spring in the Singer 12 design. There is an adjuster on the back of the head to regulate the strength of this spring.
However, the effect is not the same as on Singer 12 based machines. In this design, the spring pulls the lever down rather than pushes it up, but the adjuster for spring strength is continuous, so that the spring can be set to push the take-up lever up instead of pulling it down. This opens a whole new dimension to thread tension adjustment.
If the spring is set to neutral so that it doesn’t pull the lever down to the disk at its lowest point, then the lever won’t go down when the machine is not threaded. But during sewing, the upper thread pulls the lever down anyway. If the spring is set to push the lever up, the lever offers resistance to the pull of the thread. This mimics to a certain extent the adjustment of the range of motion in machines based on Singer 12. It is not the same, but it is similar, and completely against the rules, I’m sure. 😉
The take-up lever range is only 3.6cm, and there is no pause in movement on the way down. The lever reaches the lowest point together with the needle, then lingers at the bottom for a little longer before going up. So the lever only starts pulling up the thread after the shuttle completely clears the loop.
This is the adjuster on the back of the Defender’s head. The spring affects the take-up lever, and it can be wound or unwound by fixing the disk at a different point. The position in the photo is neutral – it’s as if the spring was not there. Rotate the disk clockwise to wind the spring and make it pull the take-up lever down to the cam; rotate the disk anti-clockwise to unwind the spring and make the spring push the lever up away from the cam. You can see from the dirt marks that the “normal” setting is one notch clockwise pulling the lever towards the cam.
Trying to sew on the neutral setting or one notch clockwise, gives a very uneven stitch on fine jersey, so I quickly gave up on that. Two notches clockwise gives a lovely even stitch, with 50% stretch at 21 SPI. At 36 SPI the stitching still looks even and the stretch increases to 90%.
Two notches anti-clockwise and we have enormous take-up strength, so I have to balance the stitch at a high tension which also calls for increased foot pressure. Only 20% stretch at 21 SPI but a very nice stitch for thick or heavy materials.
One notch anti-clockwise gives a very neat stitch with 30% stretch. Increasing to 36 SPI, and I have 75% stretch although some stitches are missed and the appearance is not so neat any longer.
So the best setting for fine jersey appears to be two notches clockwise (winding the spring to pull the take-up lever down to the cam).
And the winner is…
That depends on what you’re sewing! 😲 Yes, I tricked you there. The tests for stretch that I reported above, were all done with the same benchmark jersey – without lycra. This is a very important point. Although such fine jersey is the most difficult for getting a flexible stitch, lycra messes with your sewing in its own way. Jersey with lycra has more bulk, so stretch percentages increase across the board if the machine in question is able to handle this super elastic material. You need wide feed dogs and a good grip, and my Singer 48K grinds to a halt on lycra. Jones TS does well, but Veritas starts stretching out the jersey too much and gets uneven stitches – not enough grip. Serata all of a sudden produces a really flexible stitch with 100% stretch at 40 SPI without the slightest stretching out of the material. Jones Family TS, Mundlos 100 and the Defender perform steadily – increased stretch across the board, neat stitching, no tantrums.
These three machines are therefore the winners, right? For now, yes. But then I’m sure I’ll be sewing some other modern wondercloth and one of the other machines will come on top in that competition. I found it’s best to keep several machines with different properties so that there’s always a winner for every badly behaving material.