Reliably and repeatedly grading the color of diamonds

[Richard Sherwood]

Date: 11/19/2005 4:58:53 PM
Author: diamondsbylauren
Hi everyone-

I'm definately out of the loop here- could someone clear up a few questions I have?

1) I used a machine- at least 10 years ago- I believe it cas called a 'Colorimeter'. It printed out a label showing the color- even where it fell within the particular color grade.

Is that related in any way to Marty's machine?

Hi Dave. Don't beat yourself up too much for being "out of the loop". We realize that you New Yorkers are so busy being in the "center of the universe" that its hard to keep up on what's happening in the rest of the galaxy.

The machine you used was most likely the Gran Colorimeter, a handy machine that is fairly accurate operating within its boundaries.

In performance, you could liken the G.C. to the $5000 Sarin "BrilliantEye", while the SAS is more like a turbo-charged $20000 Sarin "DiaMension" in it's area of abilities and performance. The BrilliantEye is adequate in proportion analysis, while the DiaMension is superb. The Gran Colorimeter is adequate in its color analysis, while the SAS-2000 is superb.

Marty's webpage gives the best description of the SAS-2000, at the following address:
http://www.gis.net/~adamas//sas2000.html

I feel bad hi-jacking Dave's thread for a description of the SAS's capabilities, but you asked and under the circumstances Dave (Atlas) might appreciate the chance to do further penance and rebalance his karma after disparaging Marty's machine as "another way to do color grading badly".

A synopsis of the info on Marty's webpage follows:

The SAS2000 combines a state of the art, dual channel fiber optic spectrophotometer with an approximate optical resolution of 1.5nm. Ask about the optical resolution of single channel , single purpose "Full Spectrum" colorimeters now being sold.

The SAS2000 quantifies the 400nm to 950nm band in approximately 0.3nm intervals, approximately 8 times finer than the next closest single purpose colorimeter. The SAS2000 is a DUAL channel system, accurately tracking with the reference channel any variations in the illumination source spectra, in contrast to the single channel colorimeters.

Because of intellectual property copyright protection, no other system can legally offer the SAS2000's unique Diamond Quality Analysis reports, which show the transmittance spectra of the subject diamond along with the GIA certified diamond master stones transmittance spectra used as the scientific basis for color grade estimation.

The SAS2000 is a high resolution system which utilizes GIA/GTL graded diamond master stones for its color grading calibration and a numerical grading system (0-10) consistent with the AGS colorimeter scale reading, based on the 1950's Shipley Colorimeter. The user can add his masters to the system calibration to extend the color grading range of the system or create a completely separate calibration, for instance, using HRD masters instead of those graded by GIA, a recent software improvement.

A UV blocking filter enables the SAS to color grade diamonds without the influence of UV fluorescence.

The diamond grading capabilities is but one aspect of the SAS-2000. It also is a powerful spectrophotometer in the analysis of colored stones, treatment and synthesis detection of irradiated and HPHT treated and synthesized diamonds, etc, etc.

I'm trying to talk Marty into developing a color grading system for fancy colored diamonds. The machine has the capabilities, but the process would require extensive research with hundreds of sample stones. Perhaps you could help in that area, Dave (of Diamonds by Lauren). I could come up to New York sometime for a couple days and run all your GIA certed fancies through the SAS, building up a database from which to develop a machine grading system for fancies.

What do you think?

 

[Richard Sherwood]

Thanks for the offer to throw your Gran Colorimeter into the fray, Neil.

It would be a fun comparison study, wouldn't it?

Could be some surprises come out of it as well. I remember recently reading a comprehensive review of motorcycle helmets, where a consumers group actually bought just about every helmet on the market, and then proceeded to submit them to all kinds of real life abuse.

The helmets ranged from $80 to $600 in price. Guess which one outperformed all of them?

The $80 helmet...

I called them up and ordered three. Guess what. They are so swamped with back orders because of that analysis that it will be several weeks before they can get the helmets to me.

I thought that was interesting. People really do pay attention to these comparitive studies.

 

[Serg]

re:I''m trying to talk Marty into developing a color grading system for fancy colored diamonds

One transmittance spectrum is not enough for grade fancy colored diamonds. This task is different from grade D-L diamonds and much more difficult.

 

[Richard Sherwood]

Date: 11/20/2005 2:45:15 AM
Author: Serg
re:I'm trying to talk Marty into developing a color grading system for fancy colored diamonds

One transmittance spectrum is not enough for grade fancy colored diamonds. This task is different from grade D-L diamonds and much more difficult.

True. A much more difficult task. I figured we'd start with type 1A yellow diamonds, as 95+% of all fancy yellows are type 1A.

 

[adamasgem]

Date: 11/20/2005 2:45:15 AM
Author: Serg

re:I''m trying to talk Marty into developing a color grading system for fancy colored diamonds


One transmittance spectrum is not enough for grade fancy colored diamonds. This task is different from grade D-L diamonds and much more difficult.

Sergey is right there..
because of..
1) Becuase of the key color concept employed in fancys
2) the non homogeneity of the color concentration..

A single transmittance spectra will hone in one the bulk hue, but the value and chroma will be lower than what you see in the face up position.. the tools are there but the database isn''t for that.. and it is a "moving target" relative to GIA grading

However, you can do an effective correlation analysis, if you have enough "blessed" graded stones. I have played with this.

It is similar to the rough to polished issue which I know how to attack but just need to build the database..

Rich..

Send me spectra, and scan a copy of each "cert"

I''ll talk to you about how many you need to you need to do it.. I have a few fancy colored diamond dealers who use it that way, but the info is "valuable" and they really don''t want to share it.. Depending on the shape of the stone, it may require multiple spectra, but it is doable

 

[Serg]

re:A single transmittance spectra will hone in one the bulk hue, but the value and chroma will be lower than what you see in the face up position..

The main problem

Most fancy color diamond have big clusters with quite different value and chroma )& ( GIA did not publish rules: When could you use only one cluster( and Which for grade diamond).

Second problem ( much more less) : (Even for one hue ( Grade is in 2D space( standard color grade for yellow line is 1D)) &( GIA master fancy color diamond is absent)= There is necessary a lot of samples. But is possible.

First problem could have not good solution
I think There is not difficult develop better system for grade fancy color than GIA, but develop correct system with good correlation with GIA may be impossible. It is

 

[oldminer]

Serg and all; I know that publishing data is something that has been requested many times. It will happen over the coming months and more people take on the testing of the technology in their own labs. Right now, a place that is well known and respected has a grading machine on loan and is running several thousand GIA graded diamonds for color accuracy. If and when they accept this technology, there will be some sort of major announcement when they feel ready to present it. Of course, since they will have bought into the use of the device, they will also be accused of some bias. Its tough to be on the front of change, but way better than being last.

Please forgive me for any unintended discourtesy to any of you responding. I did not mean to imply anyone was lying because of their vested interests in their chosen devices. The same goes for me, too. I am not lying because of the perception of a vested interest. I thought this is a good topic, something not as often discussed as light performance. It has had good readership and some excellent responses. It helps me to judge what the consensus is on where we are today in color grading.

 

[mackley]

Dave,
You are a respected member of the gemstone community. You have probably mastered the comparisons of a master diamonds against another diamond to be graded. I firmly believe there are many graders (including the labs) that have not mastered this technique. My experience with lab grading is ''mixed at best'' on agreeing to the color grade provided. Even one of the AGS CG''s in my city verbally admitted to me he grades the color of a diamond out of his head. He has no masters. There will always be disagreement in color. But, we should all strive for acuracy.
I''m also an owner (serial number 3) and a user in the SAS2000 equipment. I was a believer before I purchased it. And I find it amusing for anyone purchasing one and not being a believer. That''s a lot of money just to throw away.
I use mine daily and with references to my master diamonds. So, before I put down a color grade for a diamond, it has had two opinions . . . machine and masters. Do I always agree with the labs on color? The answer is no. Most usually, Usually, the color grade I record on a diamond, if it is different than the lab, it is lower. But I do use both master comparisons and SAS2000. Would I purchase one again? The answer is yes. If something better comes along, I might consider that as well. Three heads would be better than one or two.
Marty Haske the maker of this equipment is eons ahead of my understanding in gemology and what the SAS2000 will do. He is sometimes difficult to communicate with from time to time. Since I can''t seem to get to his level, he has to come down to mine and sometimes we just don''t meet on any level. Marty, there is no knife here. I don''t think I have stated anything that Marty doesn''t know. But that doesn''t keep me from wanting to provide the best for my clients. I will continue to use the SAS2000 as well as my masters. I will continue to have dialog with Marty. I will continue to provide the best for my clients.
Thanks for allowing me to vent.
Kindest Regards,
Joe Mackley

 

[oldminer]

Joe: I pretty much do the same as you. We do the best we can and that is all anyone could ask. I probably would buy the SAS all over again just for the spectroscope readings. It gives color grades that I can appreciate, but too often cannot use. It creates no problem used as an adviser of grading. Probably Marty, yourself and others use the device way better than I know how to use it. Like you, Marty must come down to my level to instruct me on its use.

Venting is okay. I do understand the different positions people have on this. At least we all have thought about where we stand and most of us do the right thing. Just think of the many who have no idea of the communication taking place here. Many of them are important in the business, yet don''t have a clue. Sort of like grading diamonds without master stones.

 

[Richard Sherwood]

Date: 11/20/2005 8:03:26 AM
Author: oldminer
Right now, a place that is well known and respected has a grading machine on loan and is running several thousand GIA graded diamonds for color accuracy. If and when they accept this technology, there will be some sort of major announcement when they feel ready to present it. Of course, since they will have bought into the use of the device, they will also be accused of some bias.

All the more reason to do a blind comparison such as I suggested. It would be a simple and easy thing to do, dispelling any uncertainty as to which machines grade well, and which machines grade poorly.

PriceScope is the perfect venue to do this through, as they have already broken ground with the GIA-AGS-EGL comparison, which got attention throughout the industry. I'm sure industry members would be just as interested to see the results of this comparison.

What do you say Dave? A simple and expedient way to prove your claims. It could be the best thing that ever happened to ImaGem.

 

[oldminer]

Once ImaGem is ready to go for such testing, it will be done whether they want to or not. Everyone would love to see the test happen and I am sure people will want to try it out. There is no fear on my part about a fair result here as the previous testing of the large labs was quite unbiased. It shows it can be done fairly.

A few weeks more and I think the test of color grading''s various technologiers will be part of Pricescope history firsts. It will make industry news, too.

 

[Richard Sherwood]

Date: 11/20/2005 1:40:30 PM
Author: oldminer
Once ImaGem is ready to go for such testing, it will be done whether they want to or not. Everyone would love to see the test happen and I am sure people will want to try it out. There is no fear on my part about a fair result here as the previous testing of the large labs was quite unbiased. It shows it can be done fairly.

A few weeks more and I think the test of color grading's various technologiers will be part of Pricescope history firsts. It will make industry news, too.

Dave, I think you're missing the point. You have made a claim that another man's technology by which he makes his living is only "another example of color grading badly", and inferred that the ImaGem technology is superior.

A pretty bold statement which you don't seem to be backing down on or apologizing for (although you have ameliorated your tone somewhat in the face of several respected users who have stated that their experience is contrary to your observation).

The least you can do after making such a statement is back up your claim. The testing done by the facility you mentioned previously has nothing to do with a comparison between the SAS-2000 and the Imagem (or the Gran Colorimeter). Here is your chance to conclusively back up your assertion, in a simple and expedient manner.

It would be even cheaper to do than the GIA-AGS-EGL comparison. The vendors already have certed stones, which I'm sure several would donate for a short amount of time to have graded. The analysis would be done on a blind basis, with the tester not knowing the certed color grade. They would just run the analysis, print out the data, and submit it to Leonid.

I'm assuming you would donate your time at no charge, as would I (and Neil I'm sure as well). The only costs involved would be shipping costs.

Mr. Agarwal, I suspect you are lurking in on this conversation. If you are the hangup, I would ask that you give Dave your permission to participate in such a comparison. So many claims have been made regarding this machine with virtually no published analysis of its capabilities. I'm sure many are as tired of it as I am.

Dave, at the least I think a retraction of your statement regarding the SAS-2000 is in order if you're not willing to back it up with facts.

 

[strmrdr]

This is what id like to see:
Dave A.
Richard
Neil

Vendor who allows there diamonds to be used sends it out keeping the cert.
The 3 appraisers grade the stones using first master stones then the machines.
If a stone is border line in their opinion they should say so.

Stones no flor and some with med and strong should be used.
Gia and ags certs at the min. should be represented.

Submit both results to who ever is going to coordinate and write the story.
That person then publishes the result with the labs grading.
Up until that time no-one involved in getting the results should know the lab grade or what the other appraisers/machines assigned it.

If any stone should range more than one grade over all the tests then an opertunity for an explaination should be allowed to be given by the person doing the grading if it was a human grade or someone that is an expert on the machine if its a machine grade for the highest and lowest grade the stone recieved.

 

[Richard Sherwood]

I agree with everything Storm proposes except that I think the stones should be machine graded only, as that is the point of the comparison. How well the machines perform.

 

[strmrdr]

Date: 11/20/2005 3:50:54 PM
Author: Richard Sherwood
I agree with everything Storm proposes except that I think the stones should be machine graded only, as that is the point of the comparison. How well the machines perform.

how the machine grading compares to human grading is even more important than how the machines compare to each other is my thinking on it.
kinda kill 2 turkeys with one swing :}

btw i did go back and edit so make sure you re-read it.
The explaination part wasnt clear.

 

[strmrdr]

want to add that I think the appraiser should be id''ed as A,B,C and not by name.
That will provide some protection in case of a disputed score.

My worry with just the 3 machines and the Lab is what if the machines all 3 give a stone a G but gia says its an H which is right?
Having the appraisers do human grading also gives another insight into what color the stone actualy is.

 

[belle]

...reliably...repeatedly... it''s all rhetorical until we see some evidence.

 

[Richard Sherwood]

Date: 11/20/2005 3:59:49 PM
Author: strmrdr

Date: 11/20/2005 3:50:54 PM

Author: Richard Sherwood

I agree with everything Storm proposes except that I think the stones should be machine graded only, as that is the point of the comparison. How well the machines perform.

how the machine grading compares to human grading is even more important than how the machines compare to each other is my thinking on it.

The cert would be the human grading comparison. GIA & AGS are recognized as the benchmarks in industry grading, so they should suffice for the human factor.

 

[denverappraiser]

Date: 11/20/2005 4:09:45 PM
Author: strmrdr

want to add that I think the appraiser should be id''ed as A,B,C and not by name.
That will provide some protection in case of a disputed score.

I don''t count the Gran as especially reliable. It is useful in a rather narrow set of circumstances but I make no claim that it''s anything like the callibre of the SAS or the Imagem. I''m happy to use it to participate in the test but I certainly don''t want to be identified as ''the guy with the Gran'' in a study against such august competition. Me? I''m saving up to by a SAS2000.

By the way, I have a neighbor who also has a Gran and I''m sure he would be happy to participate as well and we have two different models.

Neil Beaty
GG(GIA) ISA NAJA
Professional Appraisals in Denver

 

[oldminer]

The testing will be done! I just made a long reply and it vanished into cyberspace....darn. When the substantial independent color grading tests are completed in a few weeks, we will proceed to agree on how to do the testing in a fair and unbiased way.

We''ll do a good test with all devices and independent graders. Dr. Aggarwal is travelling in India and never asked me to post this thread..

Marty knows I use the SAS2000 as I best see fit to use it and that I do not mean to stab him in the back. I have owned 2 different Gran''s, an Austron and the SAS2000. The other color machines are gone and Marty''s in still in use. Why, because we find it works. I just don''t use it for final diamond color grading.

 

[Richard Sherwood]

Date: 11/20/2005 6:35:29 PM
Author: oldminer
The testing will be done!

Excellent.

As Pharoah used to say, "So let it be written...so let it be done."

 

[RockDoc]

I have a Gran 3000.


I do think however that a human eye grading with comment about cut would be even more interesting, as I think certainly we''ll all agree that cut influences the color grading, perhaps more significantly face up.


I''ll participate in this study as well!


Rockdoc

 

[strmrdr]

kewl rockdoc glad to see you in on it.

To be perfectly honest Iv been wanting to see the result of an appraisal of the same diamond by all the PS appraisers to see for ourselves how repeatable human grading is accross top appraisers with the same set up as the lab grading study.
Im hoping that it will be done in conjunction with the machine study.
The diamonds are going to be there so its the perfect opertunity.
It could even be published as a seperate study.
Id like to see clarity taken into account on this part of the survey.

Im fine with the appraisers being identified as a,b,c,d and not by name if someone is worried about a grade disagreeing and looking bad.

thats my one last push for it to happen :}

 

[adamasgem]

I hate to burst anyone's bubble on the "study" everyone is talking about in good faith, but I feel to be really meaningful, it would have to include a statistically significant sample set for each color grade.

In my inital comment to Dave, I quoted fact that our yardsticks are not all the same length..

1) The best that GIA has ever done, to my knowledge, is quantized to 10% of a color grade range for masters, and these are for "masters"

2) The data I have, based on what I was gien by GIA, indicates (on a 17 stone sample set) that any two master stones from two different sets, having the same nominal grade from GIA, have a 30% or so probability of being a third of a grade or more apart in colro space, with some very dubious absolute placement once one gets over k-l-m color grade range, and very few appraisers know where their "masters" really lay in color space. GIA HAS THIS DATA!!!!!!!

3) Data that I have taken, indicates that fluorescence plays a major part in the grading of standard (non masters) diamonds, again see http://www.adamasgem.com/giafluor Grading technique influences the amount of UV enering the problem (when GIA stated originally stated for 50 years there shouldn't be any) . For example we can see one to two color grade shifts with medium blue, two to three with strong blue and three to four with very strong blue fluor. GIA "none" actually means that it can have some, ala Bill Clinton

4) Bruted girdle masters are useless for visual color grading unless they are clean, by being boiled in acid. I have seen "diamond masters" form some appraisers (not on this thread) which were two or three color grades off. Really BAD, and they had their tails between their legs when they had them boiled and cleaned.

5) Color grading space is not uniform in "width" for each "GIA grade" and this can effect the perceived accuracy of any testing.

6) Each color grade is really a "range" in color space, i.e. two true G's could be really one whole color grade apart, "high G" or a "low G", and all we know from a LARGE number of GIA regualr papered samples, if we assume uniform distribtion, is that the mean of the sample set would be in the middle of the range with o standard deviation of about 0.28 of a color grade range. in reality, the distribution of any reference sample is probably skewed, as the terue distriibution of "color" probably resembles the tail of a gaussian distribution. Dave's data that supposedly is being run on thousands of stones, will be intersting to see, if it isn't quantized to one "whole" grade.
Realistically, there are more "low Gs" than "High G's" in truth, and this will effect the study.

So a lot a homework has to be done before everyone starts spinning their tails, and I believe that you need thousands of samples for any tests like what are being proposed, to be really meaningful.

But that is just my( humble) opinion

 

[Regular Guy]

Date: 11/20/2005 11:51:37 PM
Author: adamasgem

So a lot a homework has to be done before everyone starts spinning their tails, and I believe that you need thousands of samples for any tests like what are being proposed, to be really meaningful.

I could be proven wrong, but to achieve statistical validity, for a given color, once your other paramaters would be set up, a number of nearer 40 should be sufficient, or so I''ve been told. Someone who does research methods as a matter of course could certainly clarify.

 

[adamasgem]

Date: 11/21/2005 12:03:16 AM
Author: Regular Guy

Date: 11/20/2005 11:51:37 PM
Author: adamasgem

So a lot a homework has to be done before everyone starts spinning their tails, and I believe that you need thousands of samples for any tests like what are being proposed, to be really meaningful.

I could be proven wrong, but to achieve statistical validity, for a given color, once your other paramaters would be set up, a number of nearer 40 should be sufficient, or so I've been told. Someone who does research methods as a matter of course could certainly clarify.

It is a multi variate problem, not univatiate, in my technical opinion, and to achieve overall test validity, you are going to need more than the "large sample" univariate number, the number I have always used as a breakpoint is 31, where you don't gain that much more information or accuaracy by adding another sample like the number 40. That is a typical noise reduction problem based on the normal distribution statistics, but we are really not dealing with the normal distribution function, as I thought I made clear. Any good theoretical mathematicians out there want to look at that problem in dealing with the actual non normal distributions and degrees of freedom.

What is being proposed, along with the non-normal distributions and quantizaions involved, suggest to me the use of F-Distribution multivariate statistics to deterine how many more samples to get the same level of accuracy as your suggested 40 samples based on the the typical "student T" distributional statistics for gaussian distributed data sets.

But the BIG thought problem involved is to define the actual number of degrees of freedom created or involved in the proposed test scenario as we add graders, "master sets", machines, quantization, etc for a level of "accuracy" of the "truth" desired (i.e. are two graders (machines) statistically the same when they are using different length yardsticks) to be defined for the proposed test scenario, some of which are suggested in my previous post.

 

[Garry H (Cut Nut)]

Date: 11/20/2005 11:51:37 PM
Author: adamasgem
But that is just my( humble) opinion

Well Marty, it is great to see that you have been reading up before your first "Fireinds and Influence" class which begins Monday. Well done

Actually, seriously, I agree with Marty - I have been unusually quiet about this survey idea because it is complex.

You have yellowish, brownish, greyish, pinkish, greenish etc to contend with - and many different effects from shapes, faceting types and LXW options. Fluoro is a study all by itself.

Marty, let me put the shoe on your foot; what would be the best way to conduct a simple test / survey that was able to tell the viability of GRAN, SAS and IMAGEM, say based on something like GIA or AGS?

 

[adamasgem]

Date: 11/21/2005 1:39:12 AM
Author: Garry H (Cut Nut)

Date: 11/20/2005 11:51:37 PM
Author: adamasgem
But that is just my( humble) opinion

Well Marty, it is great to see that you have been reading up before your first ''Fireinds and Influence'' class which begins Monday. Well done

Actually, seriously, I agree with Marty - I have been unusually quiet about this survey idea because it is complex.

You have yellowish, brownish, greyish, pinkish, greenish etc to contend with - and many different effects from shapes, faceting types and LXW options. Fluoro is a study all by itself.

Marty, let me put the shoe on your foot; what would be the best way to conduct a simple test / survey that was able to tell the viability of GRAN, SAS and IMAGEM, say based on something like GIA or AGS?

Send money Garry.. I have limited time and resources.. I''m not funded like Imagem or am printing money like you are in your store

About a two years ago, I looked in a major database of what was "avaialble for sale" in GIA and AGS "certed" goods, limiting my search to round brilliants 1.00 to 1.49cts with limitations inclor/clarity There were over 26000 stones in this class and the distribution for color and clarity used were as shown in the attached picture.

I talked with Leonid and he said he would program an automatic similar search, such that the data could be used generally and we would have sort of up to date information for any study. The attached picture took me a lot of time to construct manually, and I''m sharing it here. One of the problems is to eliminate virtual inventories, where many dealers are trying to sell the same stone, in order to get a truer perspective of what is actually on the market or in the ground..

Reaiize that lower color grades are typically not certed or "graded" by labs other than GIA or AGS and the picture may change with size or shape, so we are look at the tail end of some distribution. What we don''t know is the distribution of the color within the grade range,

Not only is this type of data usefull n determining the relative "weighting" that should be applied to any study but I also included a picture of the relative rarety of higher clarity grades, which might

But it also shows a distributional trend for certan colors, and may imply the physical dstribution of true colors in the trade in general. That is one part of the puzzle.

 

[Garry H (Cut Nut)]

We did exactly the same in selecting the grades for the GIA AGS EGL USA study Marty.

3.1. Relative comparison of the Clarity and Color grades

Table 1.a. Carat, color and clarity of the diamonds used in the survey - we focused on avaliability and also where we believe the greatest internet business is done - this table is gibberish without formating - it can be seen here http://grading.pricescope.com/color-clarity.aspx
--------------------------------------------------------------------------------

No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Carat 0.74 0.78 1.08 1.11 0.62 0.60 0.62 0.62 0.71 0.70 0.79 0.71 0.75 1.13 1.02 1.23 0.70
Color
GIA F F H G I I G H H H E E H J E G H
AGS F F G G I H G H H n/a E E H I F G H
EGL E E G F H H F G G G E E G I F G H
Clarity
GIA SI1 VS2 SI2 SI2 VS1 VVS2 VS1 VS2 SI2 VS1 SI1 SI1 SI1 SI2 VVS2 VS1 VS2
AGS SI2 VS2 SI2 SI2 VS2 VVS2 VS2 SI1 SI2 n/a SI2 SI2 SI1 SI1 VS1 VS2 SI1
EGL SI1 VS2 SI2 SI2 VS1 VS1 VS2 VS2 SI2 VS2 SI1 SI2 SI1 SI1 VVS2 VS2 SI1
--------------------------------------------------------------------------------

But would you be better to focus on say all the stones from 1 color?
What would be the minimum? 10, 100, 1000?
etc etc

 

[adamasgem]

Date: 11/21/2005 2:45:30 AM
Author: Garry H (Cut Nut)

We did exactly the same in selecting the grades for the GIA AGS EGL USA study Marty.

3.1. Relative comparison of the Clarity and Color grades

Table 1.a. Carat, color and clarity of the diamonds used in the survey - we focused on avaliability and also where we believe the greatest internet business is done - this table is gibberish without formating - it can be seen here http://grading.pricescope.com/color-clarity.aspx

--------------------------------------------------------------------------------

No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Carat 0.74 0.78 1.08 1.11 0.62 0.60 0.62 0.62 0.71 0.70 0.79 0.71 0.75 1.13 1.02 1.23 0.70
Color
GIA F F H G I I G H H H E E H J E G H
AGS F F G G I H G H H n/a E E H I F G H
EGL E E G F H H F G G G E E G I F G H
Clarity
GIA SI1 VS2 SI2 SI2 VS1 VVS2 VS1 VS2 SI2 VS1 SI1 SI1 SI1 SI2 VVS2 VS1 VS2
AGS SI2 VS2 SI2 SI2 VS2 VVS2 VS2 SI1 SI2 n/a SI2 SI2 SI1 SI1 VS1 VS2 SI1
EGL SI1 VS2 SI2 SI2 VS1 VS1 VS2 VS2 SI2 VS2 SI1 SI2 SI1 SI1 VVS2 VS2 SI1

--------------------------------------------------------------------------------

But would you be better to focus on say all the stones from 1 color?
What would be the minimum? 10, 100, 1000?
etc etc

Garry You clearly don''t understand.. What you did is different than what I show. I want to know the true statistical distribution of the real world X size stones by color..