# Split a Sphere - odd results

I’ve got a 14’ sphere that I’ve divided into 18 equal panels, no problem there.

When I next go to split the individual panels using a circle that runs along the equator of the panels, for some reason it gives me 3 surfaces.

If I split the panels using a plane running through the equator it splits into 2 surfaces as expected.

Here’s a closer view of the errant circle split:

I’ve split a lot of things with other things over the years and never encountered this behavior.

I think is related to a projection problem. if you use a surface, it works by the intersection but if you use a curve it uses the projection and sometimes it fails.

Hi @edhenry70,

Could you post a Rhino file that exhibits this behaviour, but undone to just before the circle split? I can’t reproduce the behaviour in Rhino 6: are you using 5?

Regards
Jeremy

I’ll post a file on Monday. I’m using Rhino 6, but the file was originally Rhino 5.
I thought maybe it was tolerance settings, but I’m working in feet with .001 (1,000th of a foot?) tolerance. That small strip in the middle was something like 1.5" wide.
I restarted Rhino because… sometimes that helps when things get strange.
Anyway, more on Monday (+ file).

In the attached file, if you split the sphere at the equator using the split circle (on layer of same name), it splits the sphere in half at the equator as expected.
If you instead split the sphere first with the split planes vertical (resulting in something like a sliced orange) and then try to split those resulting slices with the split circle it does the thing I described in the original post. Splitting any of these with the split plane horizontal works as expected.
2019 08 19 14 LED Sphere - Half and Half - split test.3dm (172.1 KB)

Hello - your circle is not quite on the surface of the sphere. If you `Pull` the circle to the sphere first, then split (or just split with a plane or an isocurve) it will all work.

-Pascal

3 Likes

Hunh… I think they’re coincident with each other?
Looking at the circle and sphere properties:

Circle
start = (7’-0",-0",0")
center = (0",0",0")

Sphere:
start = (0",0",-7’-0")
end = (0",0",7’-0")
center = (0",0",0")

If both are at 0,0,0 and 7’ radius… they should be overlapping, no?

As an experiment, I redrew the sphere and the circle, both starting at 0,0,0 and 7’ radius, and the circle splits the sphere in half as expected.

The Properties units are rounded based on Document settings, much like annotation styles. Look in Properties > Units, display precision.

Hello - I don’t know how you got there but they are very slightly off - use the `What` command on the circle and then `Pull` the circle to the sphere and check it with `What`, or `CrvDeviation` - slightly but definitely off.

-Pascal

Hi @edhenry70,

If you change the unit display to decimal, you will see that the sphere’s radius is 7.00000, but the circle’s is only 6.99940 (however the difference is within the document’s absolute tolerance setting).

When you select all the segments at once, they are being split by projecting the 6.99940R circle vertically onto the spherical surfaces, giving the three-part splits you report. However, if you only select a single segment then it is split at the equator, which I would think is the expected behaviour for items in tolerance, and consistent with the unsegmented sphere split using the same circle.

@pascal, if inconsistency is a sign of a bug then I think we do have a bug here.

Regards
Jeremy