Where two survey lines cross, you should compare the two values. Since they image the same point, they should have the same value. If they do not, at least one of the measurements is in error. For much of the historic data collection done before GPS availability, the positions of the data could be subject to large errors. Crossover analysis could adjust positions over the course of a cruise to minimize errors and better distribute them. Current technology has mostly removed the positional errors, but crossovers still provide important verification that the instrument was working, and corrections were properly applied. Ideal crossovers will have perpendicular paths.
Map showing the identified targets, and the
track lines. Most of the lines are run
NWSE, but there is one perpendicular line which
allows an assessment of the data quality.
This is often called a tie line. Map query, Rectangular region to restrict the map, and then the graphs, to just a small number of points. If you are not careful with this step, you will have a hard time interpreting your data. 

This area shows two targets, and a crossover where two flight lines intersect.  
The two passes on a LAT scale.
Stats,
2D graphs, 2D graph, simple (points),
and selected the two variables. This a a projection onto a single plane of a 3D data set, and there is actually only a single intersection. 

The two passes on a LONG scale.
Stats,
2D graphs, 2D graph, simple (points)
This a a projection onto a single plane of a 3D data set, and there is actually only a single intersection. 

Blowup of just the immediate vicinity of the crossover, showing there is about a 10 nT difference between the two passes.  
3D plot, New option controls, preferred. The ability to manipulate this view might make it prefereable to the others. 
last revision 4/11/2016