Grading from body measurements pt. 3

This is part three of an ongoing discussion about N. A. Schofield's article Pattern Grading found in the Sizing in Clothing book. Part one is here, part two here. I recommend reading the previous parts of this series before reading this one.

So what were the results of Schofield's experiment? I can't reproduce the actual results here, but it was something like this.

Imagine the square is a bodice pattern piece in one size. The star is supposed to be the same pattern piece but graded to the next size. Clearly, the two shapes have no proportional relationship to each other. The problem is further compounded by a different grade for corresponding pieces.

Imagine these are front and back bodice pattern pieces. Each corresponding pattern piece was graded separately based on the measurement data for that body location. Now imagine trying to sew the front and back together. It can't be done. Schofield freely admits the difficulty in the results. Though she also believes we need to learn how to deal with new shapes in pattern pieces in order to achieve superior fit.

Schofield's experiment left me with a lot of questions. I did not understand completely why she rejected the ASTM measurement data, nor why she went back to essentially raw data. Her grading methodology left me a bit confused. The results were clearly not suitable for industry application. Superior fit is the holy grail of fashion, but I'm not convinced that grading is the entire source of the problem. Superior fit, for each individual might only be achieved on an individual basis. In this case, 3D body scanning and customized clothing is the answer, but is it practical?

I would like to see this experiment repeated. The factors that will impact additional experiments are the measurement data and grading methodology. Why not use ASTM measurement data? Why not use traditional grading methods? I always support those who are willing to test ideas and theories. This was a worthy attempt by Schofield to ask important why and how questions.

Grading from body measurements pt. 2

This is part two of an ongoing discussion about N. A. Schofield's article Pattern Grading found in the Sizing in Clothing book. Part one is here.

My initial reaction to the idea of grading from body measurements was, "Well, of course we should." And in fact, we do for children's clothing. It seemed rather obvious to me to look at children's clothing as a model. Children's sizing is based on the idea of growth, meaning that the measurement intervals between sizes are not always consistent.

Let's look at an example for a 4-6x size range.*

For sizes  4, 5, 6, 6x
Chest: 23, 24, 25, 25.5
Waist: 21.5, 22, 22.5, 23
Hip: 23.5, 24.5, 25.5, 26.5

The grade works out to be, choosing size 5 as the base size:
Chest: 1, 0, 1, 1.5
Waist: 0.5, 0, 0.5, 0.5
Hip: 1, 0, 1, 1

In this example, we have a 1" chest grade, except for size 6x which is 1.5". The waist is a 0.5" inch grade and the hip returns to a 1" grade for all sizes. Each body measurement area has it's own grade.

In women's clothing a 2" grade means that the interval change between the sizes will be 2" for chest, waist, and hips. Though even this isn't true across all brands, and you will find variations. (IMO, this is a good thing)

I don't know the history of women's sizing well enough to explain how this mode of practice came to be nor exactly why. It is clear that it does make grading, especially hand grading, much easier in practice. It is also unclear to me that grading is the source of our fitting woes. Nevertheless, it does make sense to me to go back and look at body measurements and devise a more precise grade rule.

The question then becomes, which body measurements do we use? In my children's example above, the numbers are still nice and easy to work with. The body measurements have been intentionally manipulated to be easy to work with. Raw measurement data was averaged, sorted, and studied to arrive at some numbers. Those numbers were not easy to work with, so a group of industry professionals sat down and made them that way. They modified certain measurements by about 1/8" to achieve consistency. Their modifications were rather minor and easily fall within a statistical margin of error. If you read their reasoning, it makes sense. This manipulation of measurement data for ease of use continues today in more modern measurement studies. It seems deceitful, but at the end of the day is infinitely practical. ASTM D4910 inherits this method of data handling from the measurement studies done in the 1940s, but does provide some updated measurements.

Looking at the Misses body measurement chart, ASTM D5585, it seems to be arranged and handled in the same way as the children's body measurement chart. IOW, the chart does not show a 1, 1.5, or 2 inch grade in the body measurements. It is a lot like the children's example above. There does seem to be a disconnect between measurement data and grading, at least on the surface. Individual companies will decide how to interpret and implement measurement data, and therefore their grade rules. (IMO, I think this is a good thing). And some will use a 2 inch grade, and some will not.

So what measurement data did Schofield use? She rejected the ASTM charts and created her own version of measurements derived from body measurement studies. This presented a problem because measurement studies do not always include the measurements needed for pattern making and grading. Schofield did not normalize the data, in other words make it easy to work with. Also she had to figure out how to deal with missing measurement data. I no longer have a copy of the article and can't look back, but Schofield selected certain measurements over others. How and why she handled those measurements puzzled me.

I believe Schofield's goal was to remove the idea of maintaining an ideal proportion or predictable pattern shape. She wanted to see what the body measurements really did between sizes.

Her results were almost predictable. More on that later.

*These measurements come from the withdrawn child measurement standard CS151-50. Measurements are in inches.

Grading from body measurements pt. 1

Pattern grading is the process by which new sizes are developed from an existing pattern. There are various methods or processes used to grade a pattern. These methods include slash-and-spread, shifting, and CAD. At the end of the day, each method accomplishes the same thing, a new size.

The apparel industry has received a lot of criticism for their sizing, especially of women's clothing. At it's core, sizing goes hand-in-hand with pattern grading. You have to define your sizes in order to grade a pattern. In order to grade a pattern you have to know body measurements for each size. The common grade rules for women's apparel is the 1", 1.5" and 2" grade rules used in the United States. Similar grade rules are found in Europe and the UK. The primary criticism is that these grade rules are not based on anthropometric data, or actual body measurements. Instead these grade rules are just pulled out of a hat without regard to women or their fitting needs. These arbitrary grade rules are merely for the convenience of industry.

This is the point of view taken by N. A. Schofield in her article Pattern Grading found in the Sizing in Clothing book. The goal of her research was to test the idea of creating grade rules based on actual body measurements rather than an arbitrary grade rule. There has been a lot of criticism of the industry over sizing and it is a worthy goal to research alternatives. Asking the why questions. Why does the apparel industry do things the way they do? Why do we grade women's clothing this way? Can we do it differently? I've asked a lot of these same questions as I've looked at children's clothing. When I started out, I didn't understand the why and sometimes the answer was not satisfying. I can totally get behind Schofield's motivation to try and find an answer.

And yet, I feel like I am setting up to be very critical of Schofield's research and I don't want to give the impression, as an industry professional, that even asking the questions were wrong. She was right to ask the question and to test an alternative. The results of her research are interesting and ironically (and indirectly) add support to current practices.

So here are some of Schofield's main arguments:

1. 1", 1.5", and 2" grade rules are not based on anthropometric data. Meaning it is not based on body measurements or the proportional relationships between body parts/areas. These grade rules were intended for the convenience and ease of hand grading.

2. Grade rules should be derived from body measurements. This means that grade breaks between bust, waist, and hips should not be consistent. Instead of a 34-36-38 chest measurement, we should be seeing a 34-35.5-38 (just as an example), chest measurement.

3. Size prediction and also body measurement prediction needs refinement. This idea is rather complex. Body measurement studies create a lot of raw data. In order to make sense of it, statisticians will test size prediction by using one or two body measurements. So can you predict the overall body size by using just the height or chest measurement? And if you do that, what influence does that have on other body measurements? If a person gets taller, do they also get wider? It is a complex question and not easily answered because there are so many variables. Statisticians bring order to raw measurement data so that we can organize the body measurements into sizes. They do this by averaging and, in some cases, normalizing the data so we can work with it easily. Schofield implies that we should just rely on the raw measurement data.

The ultimate goal of this study was to improve overall fit of women's apparel by basing grade rules on actual body measurements. I'll have to break up my review of this study into multiple blog entries because I have a lot to say about it. So stay tuned.

Production Systems, Sizing, and Quality Control

Having worked in the apparel industry for a while now, I sometimes have to take a step back to see how much goes into the design, production and selling of just one item. The goal is to work out all of the bugs before an item goes into production. Still, things can and do go wrong and those things can affect the size of an item. Where can things go wrong?

1. Pattern making and grading

A poor pattern or poor grading certainly affects the size. Inadequate seam allowances, missing notches. There are a lot of little things that can go wrong, though if you read the rest of the list, you can see there are a lot of other areas that can cause problems.

2. Marking

Marking is the process of laying pieces out in preparation for cutting. Many markers are now made by computer, some automatically. There are various setting that allow the marker maker to bump pieces closer together, tilt, or rotate pieces. Each of these options can affect the final outcome, so they must be used rarely.

3. Spreading

Spreading is the process of laying fabric on a cutting table in layers. Certain fabrics stretch during this process and need a certain amount of time to relax. If the style is cut without the fabric being allowed to relax, then the pieces will relax after being cut and the item will end up too small. Layers can also shift and move.

4. Cutting

Accurate cutting is the only way to ensure the item ends up as intended.

5. Sewing

Taking too big or too small of a seam allowance affects a size. Seam allowances should allow a cut off allowance because nearly all operators trim off a little bit.

6. Finishing

Finishing includes steaming or ironing. Too hot a temperature can shrink an item.

These problems can be minimized by implementing systems and policies during each phase of development and production. One company I worked for had a quality audit system that checked each step of production. Anything that had finally made it to production had been perfected, for the most part. There was that one style that shared patterns with another style. A problem came up in production where the obvious solution was to make a new pattern for the new fabric, but that never happened. Production just had to deal with it. But those situations are rare.

In any event, systems (procedures/policies) have to be implemented for development, production, and finishing to catch problems before the item is shipped to the customer..

This blog entry was inspired by the article Production systems, garment specifications, and sizing by S. P. Ashdown, L. M. Lyman-Clark, J. Smith, and S. Loker in the book Sizing in Clothing.

A t-shirt refashion maybe

I picked up this t-shirt on clearance at Wal-mart. The embroidery caught my attention and at only $5 it was worth taking a chance. The style is meant to be loose, so the t-shirt itself is boxy. There is a drawstring casing at the high hip level to cinch it in. I did get compliments when wearing this, but it was not comfortable. The fit was just too sloppy for my tastes.

So I pulled out my TNT t-shirt pattern that I made a few years ago. My initial thought was to just cut down the shirt to match the fit of my pattern. Essentially using the t-shirt as a fabric rather than a shirt.

There are a few problems that stopped me. First is the casing. I debated on cutting out the casing and just having a band on the bottom hem but I was concerned the seam would be in a weird place. The next problem was the sleeve. There really isn't enough fabric to recut the sleeve.

Looking now at the picture, I wonder if I should take in the sides and call it good? What do you think?