In theory, to cut a 350mm piece out of a longer board, I would have to mark the line at 350mm + half of blade width. So if I have a 2mm hand saw, the line would be drawn at 351mm. How does that work in practice? Should blade wiggle be taken into account? Is it even reasonable to expect such precision in home conditions?
Yes, you might want the highest precision possible - for example when you build furniture which benefits greatly from precise cuts. The most convenient way would be to draw the line that will signify the edge of the piece you want cut so that the blade cuts the line and whatever material is on the far side of the raw board. Something like this:
|the detail you want|line|remaining material| | |blade cuts here|
This is quite easily achieved with power saws even without a guide and will be more of a challenge with hand saws.
You're looking for a precision cut, with a non-precision tool.
It's better to cut the piece slightly larger than what you need, and sand/plane to the final dimensions. This will allow you to compensate for blade wiggle, blade bevel, human error, chip out, etc.
The old adage should go Measure twice, cut once, sand to fit.
I would mark the line at 350mm and cut with the edge of the saw against the line, away from the edge you measured from. That way you're not trying to keep the middle of the blade on that line, you just keep the edge on the line, and you don't need to know the width of your saws.
With a circular saw the leading edge of the cut from directly above is obscured by the saw guard anyway, so you have to watch the cut from the right hand side, which would make keeping the cut in the middle of the mark very difficult.
I always take saw blade thickness (and pencil mark thickness, for that matter) out of the equation entirely by thinking of each as having a single reference edge.
The blade, in this mental model, doesn't "take out" a blade width, it cuts an edge and leaves "slop" on the other side (doesn't matter if the blade is 1/16" wide or, in theory, 3/4" wide... what's chewed is slop and all that matters is the reference edge). Similarly, the pencil line doesn't have a width, it's merely there to locate an edge, the other side of which is slop.
When I'm using a chop saw, for instance, I'll draw my line on the material (one edge of which is my reference), then I'll sight down the edge of the blade (accounting for the actual "bite" edge of the teeth, which are often alternated right/left slightly). Depending on what I'm cutting, I might even take a tiny nip to be sure the blade is biting where I thought it would, then I'll commit with the full chop.
When you're marking and cutting, think about edges--not centers or widths. Both your marking and your cutting will be much more precise.
i wanted to clock in and put a scenario into play.
this utilizes what others have said and puts it into a clear practical example.
conventional wisdom and practice cannot be circumvented. measure TWICE and cut ONCE then SAND to form.
in the mathematics, we can take things a little further. but in practice this information is almost useless.
Given: I want a 8- piece circle to be constructed out of blocks who have been cut to a repeating angle. and if these are to be glued together without touching up and shimming the edges, accuracy is necessary. each cut has to count.
so I take 1" block stock wood and use a mitre saw. I determine the desired circumference and measure the thickness of my blade.
Lets call it a 12" diameter circle and thus a 6" radius. this results in a 37.74" circumference.
this means we need at least 37.74" of material
then lets say the blade is 1/16" thick. Then we must account for half of the thickness. so for each cut we mark a line where the block ends and then one more mark that is 1/32" further away. then we must multiply 1/32" by the number of cuts which is 8. that results in 8/32 of margin we must have added to our wood stock length in order to accomodate the blade error.
meaning that you need atleast 37.74 ~ 38" plus 8/32 ~ 1/4":
It would be good to have 38 and 1/4" of wood for a 37.74 circumference circle under these conditions.
if you were cutting with the precision of an accurate mitre. and lets say 360 divided into 8 pieces accounts for 45 degrees per block, because 360/8=45; then each side of each block has HALF of that. which means each cut will be made at 22.5 degrees.
BUTTTTTTTTTTTTTT! you have to do some geometry to re-account for that error since there will be an accumulation if these are cut repeatedly and have the intention of fitting together.
the geometry is as follows: the 22.5 degree angle is going to lose some accuracy because the blade thickness that we account for on each cut of 1/32 of an inch is, lets say 1/32" of a 1- inch side of one block-- 1/32 is roughly ~ 0.03125 or 3.125% of that side assuming 1- inch. but it is actually not 1- inch per side, the exterior and interior sides are wider and shorter respectively and have dimensions of about 4.75" Exterior and 3.75 interior respectively. under the existing condition of a 12" diameter and/ or 37.74 circumference circle.
so we take that percentage and apply it to the real inside and outside and we draw a tertiary line through the two that we've already put.
looks like: 3.125% x 4.75 and 3.125% x 3.75
it would look like 1 line, then 1 more parallel line and then 1 more slightly tangental line emitting from the short ("interior side") corner diagonally past the long side ("exterior" side).
It is tangental because 3.125% of a small number versus 3.125% of a large number has a different sloping relationship.
Then you add a fourth and final line parallel to that third line, 1/32 of an inch outside of that.
And finally, to accept and average these lines correctly, you must find the midpoint between parallel lines, and mark it with a 5th line in RED then and another 6th line in BLUE and then measure their slopes and average them and cut, you may find that the truest angle to cut at is not 22.5 but actually something like 22.6 depending on the circumstance.
sorry if the numbers are fuzzy but this is the rough idea. This helps me to think of it. but being a control freak over tiny numbers isnt going to get you the correct cut, in the end, it takes time and skill and practice. once you've found a successful means of cutting and shaping your parts, work backward and apply that method to what tools you have and reconcile it.
In the end you just have wood and a saw, so consider that you must compensate by a fraction of a degree and also compensate by a fractional distance in order to achieve the result you're looking for. blade thickness effects both.