How should I take the '1–2cm margin of error' on the size chart into account when choosing a size?

Let us look at the numbers first. For a top in size L, measurements are provided down to shoulder 55.9, chest 75.8, sleeve opening 23.5, sleeve length 63.2, and total length 77.8 centimeters^[1]. XS is also recorded by item as shoulder 43.2, chest 45.5, sleeve width 15.9, sleeve length 61.9, total length 74.4, and waist 38.5 centimeters^[2]. And below that, a notice is added stating that differences of about 1–2cm may occur depending on the measuring method^[3].

The items in the table vary depending on the product. For bottoms, size M is laid out as total length 112.0, waist 83.0, hip 60.4, front rise 34.5, thigh 41.4, and hem 26.3 centimeters^[4], while for some products only the two values of Bust and Length are presented as S 35.6 and 44.1, M 38.1 and 44.7, L 40.6 and 45.3^[5]. Thus, when weighing between two sizes, the item to use as a standard also differs by product.

When actually choosing, measuring clothes you own using the same method and comparing the same items directly is the closest approach. In general, because these actual measurements are taken with the garment laid flat, the values fluctuate if the measuring point differs even slightly or if the fabric is pulled. Therefore, if two sizes diverge by a 1cm difference in a single item, it is safer to decide based on parts that are difficult to alter later, such as total length or shoulder, rather than that item. When you cannot get a sense of length from the numbers alone, you can also look at the standard of the wearing images. The model is 189cm tall and wears M^[6].

hmmm, rather than words retracting the numbers written down, this single line is closer to a sign telling you not to bet your entire decision on 1cm of one item.