If I wanted to determine the principle stresses on the surface of a circular tube could I use a 45deg planar rectangular rosette to measure the strains [tex]\epsilon[/tex]x, [tex]\epsilon[/tex]y and [tex]\gamma[/tex]xy? I only ask because on a flat surface the gauges could be oriented along the x-axis, the y-axis and 45 deg between them, but is this still true on a curved surface? Lets say the y-axis is parallel to the axis of the tube. The x-axis is obviously perpendicular to the y-axis but this means the x-axis is tangent to the surface of the tube. So is it still valid to say I am measuring [tex]\epsilon[/tex]x ? Should I be using cylindrical coordinates in stead of x and y? Is there some other way to determine the principle stresses on the surface of a circular tube?