GIA versus AGS color grades

[stebbo]

The GIA's and AGS's color grades are respected as being accurate and on par with each other. The GIA grades under light containing UV, the AGS uses filtered. Why isn't the 'undergrading' of those AGS stones that possess significant levels of fluorescence reflected in wholesale pricing? I say undergrading because I believe AGS should also attempt to mimick practical conditions.

And then the Laboratory Grading Survey (which occured after the GIA switched from filtered lighting) has me scratching my head. It showed the two labs pretty much agreed on color with the AGS actually softer more times than not. It doesn't add up unless there was only 1 stone in 17 possessing enough fluorescence to cause a color shift.

 

[starryeyed]

The study is really interesting, but I don't think the survey took enough data points to have statistically conclusive data. Usually the magic number for sampling is 33.

I suspect somebody forgot their undergrad statistics class when undertaking the survey.....

 

[Garry H (Cut Nut)]

Date: 12/12/2006 11:06:31 PM
Author: starryeyed
The study is really interesting, but I don''t think the survey took enough data points to have statistically conclusive data. Usually the magic number for sampling is 33.

I suspect somebody forgot their undergrad statistics class when undertaking the survey.....

I thought the magic number was 69, or is it 42?

But you are right SEyed - far too few stones and not enough with strong and very strong fluoro to make a study to that level.

 

[oldminer]

ImaGem has had the opportunity to examine and quantify a substantial amount of human color grading from respected labs. I believe the statistics show somewhat less than 70% repeatability in grading, regardless of the number of graders in each lab who co-examine diamonds. There is a good degree of currently acceptable flex in the GIA system of diamond color grading so long as it is done with human eyes.

It is going to be a real challenge to offer more repeatable, more consistent and accurate machine based color grading because people's eyes won't necessarily agree with it about 30% of the time. The eyes and variance in lighting, even with supposedly well made fluorescent tubes, is going to make people think that machine grading is no better than their own grading because they will be unable to check the accuracy with their own eyes.

Scientist/gemologists will have to show the factual data on each stone in order to help prove the accuracy of the grade assigned by machine. This is going to be a very challenging period where humans will have to cede their authority in grading color to man-made devices which have better, more consistent ability to extract exact color data.

GIA and AGS labs do a very good job of grading with human eyes. They are dedicated to the task. Both labs can do better once they take a serious look at what is going to be offered as a large improvment in machine color grading. The current color grading devices have gotten betterr, but are not at the level of replacing or improving the accuracy of human eye grading.
Better equipment is a short way down the pipeline and will help the labs do a more accurate job.