The first and last stones appear to be fair mid grade stones. The middle stone lacks observer’s presence contrast; I believe it would look worse than it appears in the photo.
RD22
This stone could possibly have been improved by the manufacturer; by lengthening the lower girdle facets and shortening the star facets (while holding the other proportions constant) would dramatically improve scintillation and would reduce the darkness caused by the short lower girdle facets which are the cause of the thick dark star in the table.
A second and better possibility may also have existed. Diamonds cut to these proportions are frequently the only stone polished produced from the rough. They are called ‘makeables’ because they are made without sawing from often rounded ‘dodecahedra’ rough. Of course we can not know if it this would have been possible – but often very steep crown angle diamonds can be made from such material. If it was possible in this instance, then making the crown angle 2° steeper would have resulted in a stone that scored HCA 1.2 FIC which is an abbreviation for Firey Ideal Cut. This stone would have a 4% better yield.
 |
| DiamCalc model - GIA data | Model with new minor facets | New minor facets & +2° crown |
| Light Return Mono | 0.84 | 0.89 | 0.91 |
| Light Return Stereo | 0.83 | 0.92 | 0.93 |
| Contrast | 0.99 | 1.14 | 1.14 |
| Star facet length | 55% | 45% | 45% |
| Lower G’s | 75% | 90% | 90% |
In the article, GIA only mentions one stone as being a slight fisheye (RD37 Cat 4), but infact this stone, even though it has a small table size, would show a slight fish-eye (as would RD19 Cat 4).
RD11
If this survey had involved consumers actually purchasing diamonds, spread and apparent size would have been a very important factor in the grade setting of this and some other stones. This mid grade stone is a clear example of the difference between the institutional and the real worlds. The dark upper girdle facets in this stone (as mentioned in the article) are a result of leakage that is evident in its DiamCalc modelled ideal-scope image. Upper girdle facet leakage results in reduced ‘apparent size’ or ‘spread’ because our eyes loose the ability to distinguish the edge of the diamond; i.e. it looks smaller than its actual diameter. RD11 weighs 9% more than we would expect a nice diamond with this diameter to weigh, but a diamond with the same ‘apparent’ face up spread would about three quarters of the weight. This diamond would cost half as much again as a better cut diamond with a similar apparent size.
In my opinion ranking a diamond like this as a mid grade stone could result in retailers and consumers accusing GIA of pandering to large manufacturers and dealers, some donate generously to the GIA’s League of Honour Fund.
This diamond raises the possibility of a weakness in the lighting arrangement that GIA used. The DiamCalc ray trace shows that two facet sets are illuminated by light coming predominantly from the same light source. If we imagine placing this stone on a Brilliancescope®, with its annular or circular light source that moves up and down perpendicular to the diamonds table, we would expect the diamond to remain dull unless one of the 5 Brilliancescope® reading positions happens to fall in that region. In the DiamCalc ASET™ model, employing a variation on an Ideal-scope / Gilbertsonscope * approach, we see that the bright portion of the stone in the GIA photo are predominately pink.
  |
| The ray trace diagram shows that the center of the table, and most of the crown main facets, are both illuminated from the same direction; this shows as the red on the AGS ASET DiamCalc lighting model. The stone would appear dead if light from those directions was unavailable. (The ASET blue illumination would be largely obscured by an observers head. The green colour indicates light that has come from close to the horizon, the least likely source of bright illumination.) |
RD06
GIA’s computer model predicted lower crown angle stones would have better brilliance; naturally they needed to test this with visual observations; stones were polished especially for this purpose. The computer scientists must have been disappointed by the results. On page 225 two reasons are given for mediocre performance. Firstly they mention low values for crown height and angle and secondly they refer to the stones lack of contrast and scintillation. I believe they made the pavilion angle too shallow. (It would be good if GIA stopped referring to “Crown Height”. A diamonds crown height is determined by two factors; crown angle and table size. Crown height per se has little or no determining impact on a diamonds appearance.)
A low crown height needs to be combined with an appropriate pavilion angle. GIA predictions of high light return for RD06 (WLR 3.01) in their 1998 G&G light return study failed to account for an observers head, a fact well known and discussed adnauseum. Also well known is the head obscuration of shallow crown stones has a more deleterious impact than with other proportion sets. Because they made the pavilion too shallow f particular crown height, the stone shows what all shallow stones show – too much darkness. The darkness could have been reduced with adjustments to minor facets, but a better approach would have been to cut a deeper pavilion as in the example below (A pavilion angle in between would probably have yielded a more balanced stone). It appears that GIA have failed to heed the feedback from critics and have not made appropriate adjustments to their brilliance software.
 |
| The stone on the left is a DaimCalc model of the GIA stone. On the right is this stone with similar proportions, but 1° deeper pavilion; it scores an amazing 1.11 and 1.08 for DiamCalc light return, but rates only 0.75 for contrast (questioning the Brilliancescope® results for scintillation). |
GIA have made no mention that shallow crown angle is a durability issue. Perhaps GIA will follow AGS’s lead and downgrade shallow crown angles for durability in the finalized system?
* Al Gilbertson is now one of the GIA Cut Study team
Categories 4 and 5
GIA-GTL would be under considerable pressure from benefactors and clients who submit large quantities of diamonds for grading to be liberal one the one hand. On the other hand they are an ethical and honourable institution charged with protecting the public and promoting the industry. The second stone (RD33) in this category is so poor that it is hard to imagine the need for a lower grade?
RD37
Worthy of note: The Liddicoat depth estimation is a commonly used and GIA taught method of pavilion depth estimation. By estimating the width of the table reflection compared to the width of the table, a reckoning of the pavilion depth % is made (add 1/10th of the % to 40 = estimated pavilion depth). In this case the table reflection is 1/3rd the table width. This would lead us to believe the pavilion depth is 43.3%, but the depth is actually only 42.3%. The Rapid Sight method works with a relatively small range of table sizes and is also affected by girdle thickness and crown height. In this case a 1% over estimation of depth percentage would indicate the stone had a 40,9° pavilion; knowing the pavilion was only 40.2° might influence a buyers decisions.
Category 5
I question the validity of this grade that would apply to the worst 1% of certified diamonds based on a survey conducted on Rapnet on 11/19/2004. There were only 441 plus 1.00ct rounds with depth % greater than 69% listed out of approximately 40,000 round stones searched by the same criteria. Surely the lowest grade in a 5 grade system should be set to reflect around 10 to 20% of the goods in the market?
It would also have been instructive to see more than one example of a diamond downgraded for other than having an overly thick girdle.
The optical performance of RD39 appears to be as good, or better, than some stones in Categories 3 and 4. Aside from the extremely thick girdle (which does not have a major impact on optical performance) the crown and pavilion angles represent the steepest and deepest AGS 0 angles; was the inclusion of this stone a political choice?
Discussion
The GIA study noted during the observational survey that factors other than proportions played a role in diamond beauty. This table attempts to simplify those observeations.
| Factor | No effect | Slight effect | Noticeable reduction |
Color | D decrease in apparent brightness as saturation increases Z |
Clarity | IF to SI1 | Grade determining Clouds in SI2 | Grade determining Clouds in I1 > |
Symmetry | Excellent VG Good | Fair > reduced brightness > Poor |
Polish | Excellent VG Good | Fair> reduced brightness and fire > Poor |
Girdle | No effect on apparent brightness and fire from girdle condition |
Fluoro | No effect on apparent brightness and fire |
Factors that showed discernable effects on brightness and fire that were noted by observers during the GIA survey study.
Conclusion
There are a number of instances in the paper where references are made to diamond performance or appearance based on percentages or preconceived trade based ideas of how a diamond should look, or what proportions are considered ‘acceptable’. On page 220 they state “For example, we downgraded diamonds with pavilion angles that were very shallow or very deep because these proportions generally changed the face-up appearance of the diamond in ways that made it less desirable to experienced trade observers.”
The findings of this GIA study, and other studies and approaches to cut grading, will fundamentally change industry expectations. If part of the purpose of this GIA article is to change those preconceptions to match the findings of human observers and the realities of the laws of physics, why be considerate of preconceived trade customs and opinions? As Paul Slegers says, we should not be led by logic dictated by adages like “my child is beautiful”. This would lead to a lost opportunity to take full advantage of the laws of nature to maximise both the beauty and the yield on each and every one of those “finely crystallized carbon” creations.
Paul also notes that trade observers are educated in a certain way of looking, judging and thinking. No matter how ‘blind’ you make their observations by not giving them any parameters before judging, they can identify certain parameters and they may inevitably be influenced in their observational judgements.
This GIA observation based study used diamonds with considerably different weights and diameters; I believe it is un-reasonable to compare diamonds with noticeably different diameters. Diameters were not published, but by estimation with DiamCalc using published proportion data, they ranged from approximately 4.7mm to more than 6mm. Some of the 15 example stones included heavier diamonds that had larger spreads, and some of the deepest diamonds had the lightest weights.
There appears to be no part of the study that included apparent or observed differences in spread. Stones like RD11 in Cat. 3 and RD33 in Cat. 4 would appear significantly smaller than their measured diameters because of greater upper girdle leakage. In my experience this is a major factor in diamond desirability; it would result in diamonds of these grades being rejected in favour of stones with a lower GIA grade but larger apparent size.
For all practical purposes, GIA has designed a 4 Category grading system. The fifth Category may only apply to the worst 1% of certified diamonds. It could be argued that there is a larger percentage of overly deep uncertified ‘promotional’ grade diamonds, but one can not help wondering if this category was set to appease some sectors of production and wholesale within the industry? Does it serve retailers and consumers to propose an irrelevant fifth grade when there are stones in the fourth category that could already be considered by many as only a little better than unpolished rough diamonds? Cut quality is confusing and creates doubts in consumer’s minds. Doubts are impediments to buying; that inevitably reduce the growth of diamond jewelry as a whole. Brighter, more firey and sparkly diamonds sell themselves and keep customers coming back for more.
It would appear the basis of the GIA grading system will be the proportion data from a scan. Such a system can be described as a parametric grading system that employs look up charts similar to HCA (but with additional minor facet proportions and symmetry and polish grades). The AGS will shortly introduce a more advanced system using parametric and direct assessment techniques. Neither system appears to account for differences in diamond size and accompanying effects on appearance. Neither system is readily functional for the designing and planning of polished diamonds by the scanning software that is widely employed in the industry for rough diamonds. The manufacturing industry will discover that adapting to these new grading systems is rather difficult. OctoNus is continuing with the development of a 3D software based grading system. This system will account for diamond appearance based on size differences, the same system will work for any shaped diamond and it will function in reverse by enabling the best yield planning and the production of rough diamonds into the most attractive gem possible. Eventually this approach to cut grading could do away with the encumbrance of predetermined faceting arrangements; imagine non symmetric one of a kind diamonds with optimal beauty?
Symmetry and averaged data
It appears from the articles and public presentations in Basel and Hong Kong that the GIA system will use averaged crown and pavilion angle data. Lower symmetry grades will lead to lessor overall cut grades. There are at least three shortcomings from this approach.
Firstly consider the idea that there can be ‘sweet spots’ within a grade. If all the measured proportions are such that the stone is well within grade boundaries, then symmetry deviations may have little impact on the diamonds appearance. The GIA authors have qualified this by the fair claim that a diamond in the top grade should also meet certain crafting quality standards. But consider a diamond that is near the boundary of a proportion grade that has certain types of symmetry deviations, like for example, a slight squarish out of roundness that results in the combined effect of steeper crown facets directly over deeper pavilion angles. The additive effect of this type of small symmetry deviation on the ray paths in a diamond can result in far greater leakage and reduced light return. AGS would be able to downgrade such stones because of their combined use of parametric and direct determination using their 3 colored ASET scope.
Secondly, the GIA article mentioned that diamonds that exhibited excellent symmetry with Ideal-Scope or Hearts and Arrows viewers were not rated any higher by observers than those with lessor patterns. I suspect that their study only focused on diamonds within very good proportions, and in this respect, Peter Yantzer and I have never subscribed to the idea that Hearts and Arrows diamonds offer any better appearance to the naked eye than well cut diamonds with small symmetry deviations. Had GIA tested diamonds with the ideal-scope with the purpose that it is designed to be used for, rather than simply as a gauge of symmetry, I feel they would have discovered benefits form its usage and weaknesses in their current symmetry grading method.
 |
| These stones have a 1% out of round squarish profile, yet they would be graded as having excellent symmetry by all labs. The stone on the left - four of the eight sets of main facets have crown and pavilion angles of less than 34.9° and 40.9°, while the section that is leaking with a less well defined star, has crown and pavilion angles of 35.1° and 41.2°. The HCA scores for each set of proportions are 1.7 and 3.4 respectively. The stone on the right is the same model with a one half degree shallower pavilion; it is in a “sweet spot” with HCA scores of 0.7 and 1.4 respectively. |
Thirdly, another disadvantage of a parametric system is that it takes no account of a diamonds size. GIA will give the same grade for different sized round brilliants with equal proportions[ii] . Tom Moses acknowledged that a 1/3rd carat diamond with equivalent parameters would have a different appearance to a 5ct diamond. Better symmetry in a well cut 1/3rd carat diamond aids in producing easily discernable firey and contrasting facets. But in larger diamonds, of say 5 carat or more, some types of symmetry deviations can actually improve a diamonds appearance[iii] by increasing the frequency of flashes from an observably larger number of virtual facets, even though both stones have the same number of facets.
The current methods for grading symmetry has evolved from the basis of the capacity to assess or detect deviations. These systems can be shown to be inadequate for the current task.
Rounding errors
As noted, the GIA has applied a rounding of parameters, presumably because of the inherent accuracy of existing scanners. Rounding introduces errors. Accurate measurements of azimuth are also critical for parametric grading when considering the Indexing; Sarin and Ogi scanners appear to have limitations in this area.
Upper and Lower Girdle Facet Indexing
GIA published an article 13th Feb 2004 discussing azimuth and angular adjustments or minor facets that are achieved with the aid of ‘indexing’. The accounts of the affect of these polishing practices on diamond appearance ran counter to the widely held opinions of others in this field. For instance diamonds that have painted or built up upper girdle facets like EightStar diamonds® were said to show “little contrast except for the dark appearance of the pavilion mains beneath the table”. The current article makes no mention of how painted or dug out facets will be interpreted in a new grading system, nor of how these azimuth shifts could be detected with adequate accuracy.
[ii] I asked about this on behalf of Sergey Sivovolenko from OctoNus at the GIA Basel Gemfest 2004 presentation. Tom Moses replied that they would give the same grade to a 0.30ct diamond and a 5ct diamond if all proportions etc were the same.
[iii] Personal communication with Pol Van der Steen (DiamCad, Belgium) who is a specialist manufacturer of very large diamonds.
Future Pricing
The issue of the effect on pricing of diamonds was not mentioned in the GIA Foundation article, however I feel that this is one of the most interesting aspects from most readers perspective. This is my opinion; the market currently has a comparatively narrow band of price differentials for variations in cut quality. For instance dealer to dealer 1.00 ct round brilliant Rapaport pricing can range from ‘list’ down to 50% discount; put simply a good cut can cost twice as much as a bad cut. Compare this to Clarity and Colour Rapaport pricing; the average difference between IF and I3 is 9 times and for D to M the figure is 4 times.
What will the market do with this new information? It is likely that over time a new market pricing will evolve. During the year after GIA’s system is released I predict that the gap will widen to 2.5 times. Over a longer period 3 to 4 times (67-75% discount) could be common as poor graded goods become more difficult to sell. Consumers will become more aware of the need for a grading report with a reliable and respected cut grade.
Possible GIA terminology | Discount to the next lower grade | Short term prediction |
10% | 15% | 20% | 25% |
Excellent | 100 | 100 | 100 | 100 | 100 |
Very Good | 90 | 85 | 80 | 75 | 75 |
Good | 81 | 72 | 64 | 56 | 60 |
Fair | 72 | 61 | 51 | 42 | 50 |
Poor | 63 | 52 | 41 | 32 | 40 |
Finally, it is our opinion that the foundation of the GIA system is not built on rock hard science. This most recent article is verbose and vague in many areas. But it appears that by and large the GIA is moving forward.
Acknowledgements
The following people have contributed ideas and or reviewed and made suggestions to this review:
Paul Slegers
Yuri Shelementiev
Sergey Sivovolenko
Pol Van der Steen
Peter Yantzer
Leonid Tcharnyi has spent hours formatting and putting it here on Pricescope.
Thank you all.