1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Gia vs AGS cut research a contrast in styles.

Discussion in 'RockyTalky' started by strmrdr, May 22, 2005.

Thread Status:
Not open for further replies. Please create a new topic or request for this thread to be opened.
  1. strmrdr

    Nov 1, 2003
    by strmrdr » May 22, 2005
    It seems like to me that when GIA and AGS set out to design a cut grading system they went about it in almost totally separate ways.

    GIA primarily took a look at what was available and using mainly observation research separated them into broad categories.
    AGS on the other hand seems to have in effect said; if we were cutting the best diamonds in the world what would they be like?, then did the scientific research to find out.
    Then they gave the cutters the information to cut these diamonds.
    Nowhere is this more apparent than the AGS princess cut grades because they found that none of the current cutting styles was good enough and took them in a new direction.
    With rounds it was less revolutionary because there were already cutters cutting super high performance rounds but it is none the less a huge advance in diamond cut grading.

    What do you think am I right or wrong in how I see it?
  2. Serg

    Mar 21, 2002
    by Serg » May 22, 2005

    Hi, Strmrdr

    I Think your interpretation is not correct. Your forgot history GIA research

    re:GIA primarily took a look at what was available and using mainly observation research separated them into broad categories.

    It was second GIA attempt.
    AGS way is similar to first GIA attempt( unsuccessful).

  3. strmrdr

    Nov 1, 2003
    by strmrdr » May 22, 2005
    If Im reading your post right, and I might not be, how the current studies were done as they stand today Im right on.
    However GIA tried the same approach as AGS and abandoned it earlier?
    Iv read some on GIA''s first attempt but it happened before I became a diamond nut so im not very up on it mainly because im not interested in a failed study.
  4. Garry H (Cut Nut)

    Aug 15, 2000
    by Garry H (Cut Nut) » May 23, 2005
    storm here is a bit of history - an early working draft of an article later refined and published in the Australian Gemmologist.
    Good bed time reading.
    Quite a few bits I now would change. Not sure if some of the edits will come out.
    I started writting it Easter 1999

    ‘Modelling the appearance of the round brilliant cut diamond: Analysis of brilliance’
    Gems & Gemology. 34, 158-183.

    Garry Holloway, Melbourne

    In 1919 Marcel Tolkowsky [Recommend change to “Tolkowsky” throughout] published a treatise on the brilliance of round brilliant diamonds. He proposed a single set of proportions which became known as the ‘Ideal Cut’. No subsequent studies have improved on Tolkowsky’s crown and pavilion angle combinations, although many support larger table sizes.
    In the Fall of 1998 the GIA published a complex analysis of 20,000 proportion variations - computer generated ‘virtual’ diamonds. What was the conclusion? “…we have focused on brilliance….the main factor of diamond appearance…..” “….we found that the relationship between brilliance and the three primary proportion parameters (crown angle, pavilion angle and table size) is complex, and that there are a number of proportion combinations that yield high WLR (weighted light return) values.” The GIA is continuing the study to include fire and scintillation; “It is our opinion that any cut grading assessment devised in the absence of this broader picture is premature”.
    Mr Holloways article presents a solution he discovered a decade ago he believes is supported by the data provided in the GIA study.


    The patterns in Figure 11 in the Gems & Gemology paper by Hemphill et al. show a trend. The centre point of the oval ‘bull's-eyes’ on the seven graphs for crown angles between 30.5* and 36.5* move from the top right to the bottom left. This trend confirms my simple system for buying, unseen, diamonds with the aid of Sarin reports.
    (Please note that this is Mr. Holloway’s system, not one proposed or necessarily supported by GIA. )

    A range of inter-related crown and pavilion angles

    After discovering the relationship between crown and pavilion angles from observations of round brilliant cut diamonds with the Japanese invented ‘Firescope’ TM I attempted a two dimensional proof. I based this proof on an overly simplified concept that most light reflects twice from the pavilion facets and only passes once through crown facets (either on entry or exit). The exception is a small amount (<10 per cent for the Tolkowsky ‘ideal’ cut) of light that both enters and leaves via the table. (Our research shows that almost every ray travels through the diamond much more extensively than described here. Figure A-1 was a typical case, not an exceptional one; we have never computed whether a typical ray travels twice as far through the pavilion as the crown of the stone Surely Ilene most of the light in A-2 (not A-1) exits via the table on the 4th interception? ie enter via table, bounce off 2 pavilions and more than 80% of light exits via the table? )

    I found For each additional degree of variation from a pavilion angle of 40.75*, the crown must be two degrees (2.0*) less than 34.5*. For each degree of variation less from a pavilion angle of 40.75*, the crown must be two degrees more than 34.5*.
    The only other reference to this concept I am aware of is that made by Richard Von Sternberg, who has suggested a 1:1 relationship between crown and pavilion angles. He too has experimented with the “Firescope”tm.

    In my experience the maximum ideal pavilion angle for diamond is around 42.2*, which represents a 45 % depth. In this extreme case the round brilliant should have a shallow crown angle of 31.6*. In contrast, the minimum pavilion angle for the round brilliant diamond is around 39.7* or a depth of 41.5%. In this shallow pavilion brilliant the crown angle should be a high 36.6*. Table sizes should be small in round brilliants with shallow pavilions to ensure there is no girdle reflection or ‘fish eye’ effect. The precise limits and the quantification of this principal will need to be proven beyond reasonable doubt before it can be used by the GIA as the basis for establishing an effective and more achievable standard for ideal cut.

    From comments on P173 referring to figure 11, it is apparent the GIA were unaware of the significance of this data. After subsequent communications and meetings with the authors they are still not sure they agree with my findings.

    [I am not sure what Mr. Holloway means by this; we were certainly aware that the center moved to shallower pavilions and larger tables as the crown angle increased well before Symposium. Furthermore, I am not sure that we agree with his “significance” as it is described below.]

    This graph shows my linear standard for crown and pavilion angles and a plot from the WLR bullseye's in Figure 11, page 178 Gems and Gemology Fall 1998 (Hemphill, Reinitz, Johnson and Shigley). Note the close similarity. Note also how far off the two graphed lines the bottom left and top right corners the AGS '0' standard are. The GIA line passes through Tolkowsky's single target confirming the enduring validity of his result. (The Holloway line must pass through this point because I based my theory on the effectiveness of his standard.)
    The red line is a wild guess to provoke collaboration to establish a minimum standard. You are invited to participate in this study at ‘’.

    A graph of the data from figure 11 shows my 2:1 rule holds roughly true for a range of proportions: crown angles 30.5* and pavilion 42.75* to crown 36.6* with a 40.25º pavilion angle. The GIA data indicates a 2.5:1 relationship between ideal crown and pavilion angles. Why the graph creates an S curve at the extremities is a mystery to me?

    Interestingly, the WLR line passes through Tolkowsky’s point on the graph, which appears to confirm his chosen crown and pavilion angles. This evidence brings into question the GIA authors’ [this was simply a reference value for calculations, not something chosen as being especially desirable in any respect.]pavilion angle of 40.5* which they used as a reference value. (Note the position of "A" in figure11 is incorrect.)


    The ‘bull's-eyes’ position moves from graph to graph, indicating maximum WLR as table sizes decrease from 58% to 52% for progressively shallower crowns and deeper pavilions.

    [What is happening with table size “as pavilion deepens” is better seen in Figure 10. COMMENTS TO GIA-(Mary I find the information in figures 7,9 and 10 favours very shallow crown combinations. I would be interested in your opinion’s of the changes in lighting I suggest, and the results that might bring about? I expect this would enhance the peaks in figure 11 and support my observations from the firescope. I also suspect you will discover the peaks at around 23* crown angles will disappear. As for Figure 10, these are a series of horizontal slices through Figure 8. The verticle slices dispalyed in figure 11 reduce the somewhat constant variation provided by table size – Again I believe your lighting model will slightly alter the preference for smallish tables as far as ‘brilliance’ goes (dispersion and scintillation is another matter). In short, this data is very complex and the best way to view it will probably be found to be in oblique planes.) In fact, (for shallow crown angles) the table size at maximum WLR first increases from about 53% to 55% and then decreases again. (Sorry I do not see any evidence for this?) This is one reason that the view in figure 11 is not the whole story.] (Does “move lower” refer to the decrease in WLR value or the trend of the center of the contours to lower pavilion angle? Thanks Ilene, does the adjustment above suffice?)

    Both sides of the table size debate have valid arguments. The trade prefers spreads of +60% because the yield from rough is higher. Small tables look great to me, but spready stones sell best because they look bigger and brighter in jeweller’s display cabinets.

    The GIA’s further research will probably prove Tolkowsky’s small table size provides the best dispersion and scintillation. However, table size is set in the real world of availability. The GIA examined a sample of 67,621 stones and found a mere 76 (of the 2051 that had crown and pavilion angles close to 34.5* and 40.75*) had a 53% table. [in fact these 76 diamonds also occurred within restricted ranges of crown angle (34.0 to 34.9) and pavilion angle (40.2 to 41.3). We did not provide the total number of diamonds with 53% tables. Thanks Mary, is this better?]. Tables are usually bigger than trade buyers like; but diamonds with big tables sell well to many consumers, so who is right?

    If the table of a diamond is too large, you can readily see the reflection of the opposite side girdle. This creates a ‘fish eye’ as light leaks out the girdle. Shallower stones must have smaller tables, but as the pavilion gets deeper the table can be larger. The largest I would consider is a table of 62% for a pavilion of 42º and crown of 31º- 32*. The WLR results do not support a table this large. However stones of these proportions look very bright under a ‘firescope’ when visually assessing diamond’s brilliance and light loss.

    [I have always been curious as to why we are quoted two comparatively meaningless pieces of data— Table percent (%) and total depth percent (%). If we all agree that no one agrees on the right size for table %, thensurely the most important information on a report or certificate would be the angles of pavilion and crown facets? To quote; “The wide, gently sloping top of this graph suggests that WLR is not strongly affected as table size varies between 50% and 62%…” from Fig 6, page 173. While table size is important the relationship of the crown angle to the pavilion angle is most critical to the assessment of the beauty of a diamond.

    [with all other proportions at the reference values, not necessarily in general. (Valid point, but I doubt brilliance will be affected greatly ) ]

    My own observations, confirm the GIA’s findings of limitations in existing grading systems. Current systems are based on different tolerances of variation of crown and pavilion angles usually centered around Tolkowsky’s two angles. This allows stones with inappropriate combinations of crown and pavilion angles to be given excellent grades. Some beautiful stones with very favourable combinations are also given only fair to poor gradings.
    For example, the strict American Gem Society (AGS) ‘0’ or ‘excellent’ cut grade with a maximum crown angle of 35.8º, a deepest pavilion angle of 41.2º, and a 53% table (WLR .275), does not look as good as a ‘1’ or ‘very good’ with a shallow crown (33.6º), deep pavilion (41.3º) and a table of 59.5% (WLR .280). To add insult to injury, the lct ‘1’ spreads 0.15 mm larger and has a higher WLR as described in the GIA paper.

    In reality with the usually present dirt, soap etc. on the back of these two diamonds and the ‘1’ will still look good. In contrast, the upper girdle facets around the perimeter of the ‘0’ leak light out the back of the stone. From a meter away the ‘0’ will look about the same size as an ideal cut 0.90ct. No amount of extra dispersion or scintillation (both of which will be enhanced in the ‘0’) will make up for the loss of brilliance. Perhaps some virtual ‘dirty diamonds’ should be run through the GIA’s computer model by raising the refractive index of the virtual ‘air’ at the pavilion interface?


    Finally, the last of the 4C’s (Cut) is about to become just another line on a diamond certificate. The trade’s worst fears of complete commoditisation are being realised. Will we, the commercial sellers of diamond, be replaced by 16 year old ‘nerds’ “flipping certs” (as Martin Rapaport calls it) on the internet?

    . In the 1950s the GIA established the world’s first commercial standardised diamond grading system. By the 1980’s Martin Rapaport had established a diamond price information service for the trade. His price sheets list Carat weight ranges in grids of Colour and Clarity. Discrepancies between the classifications of buyer and sellers soon led to the wide spread use of independent grading reports for large diamonds. Yet, still there is no universally accepted quality grading system for Cut, the last frontier.

    De Beers have done an excellent job in communicating these standards to consumers around the world. However the fourth C is still not well understood. In its simple form cut refers to the shape of a diamond. As a quality factor, cut is a reflection of how much light comes back out of the top of the stone. De Beers brochures illustrate three different profiles for cut diamond. The ideal example is fine but the naive outlines and ray paths for deep and shallow profiles typify the lack of understanding of diamond proportions. (See foot note, below)

    * De Beers diamond profiles used by retail sales staff around the world have annoyed me ever since Dr Phakey, my Gemmology II lecturer, taught me to understand critical angle. For any light to escape from the shallow stone the pavilion angle must be less than 24.5º - the critical angle of diamond. Diamonds as shallow as this are very rare. I have seen such shallow cuts in old-miners cut from maccles and occasional modern triangular cuts; but never have I seen these in modern brilliants. Likewise, the deep example is wrong, for no light entering perpendicular to the table can escape unless the pavilion angle is greater than 52º. I have never seen an example of a round brilliant cut this deep.

    (Consumers ‘flex their muscles’)

    Two changes in relative values were noted in the Rapaport Diamond Report of January 1999. These indicate beauty is becoming more highly valued than rarity. Sophisticated consumer behaviour is taking over as a dominant factor in diamond values. Today, there is abundant evidence customers do not ‘trust the experts’ to determine value and quality.

    Firstly there has been a gradual reduction of the premium paid for high clarity.
    Flawless and VVS diamond values have performed poorly over the past decade. The largest price increases have been in the VS-SI clarity range as demonstrated in the ten year performance of the 1 ct diamond as illustrated in the 24 January 1999 issue of Rapaport Diamond Report .

    Second, the premium for ‘magic carat weights’ has lessened somewhat. For example, the difference in price between say a 0.90 ct and 1.00ct is down to an all time low. Martin Rapaport suggests “The message here is that the look of the stone may be more important than the carat weight and resale value”. I suspect one would find more ugly makes of, say precisely 1.00ct weight, with more good makes on either side—at say 0.90 – 0.95ct and 1.05 - 1.10ct. The GIA has the capacity to prove or disprove this theory by examining the large body of data within their ‘‘Horizon’TM (check if trademarked – I am happy to put a tm in) program.

    More than ever before consumers are making their own assessment when buying a diamond. Either they do not trust the advice they are given in stores, or they sense
    misinformation from poorly trained staff. More and more customers are buying diamonds that ‘look good’ for the price they are prepared to pay. And often the best looking diamonds in a jewellery store cost less than rarer ‘fine quality’ diamonds that simply do not ‘look good’. Perhaps we in the trade have not woken up to our customers waking up?

    The words of Al Gilbertson GG, writing in the GIA’s Alumni Magazine InFocus (W/S 1999, page 31) went some way to the heart of the matter when he stated “We are now at the crossroads of the evaluation of cut. No consumer can see the difference between a VVS1 and SI1 with the unaided eye, but now we are starting to evaluate cut at that level. Consumers are willing to pay a price difference for different clarity grades that they will never see with the unaided eye. The industry sought that standard and consumers bought it. So, the industry can set the standard for cut and, yes, the consumer will buy.”

    I agree with Al, consumers will ‘buy it’; but only because they can see the value for themselves. They will no longer accept the dogmas of self-interested groups. We the jewellery trade, are being dragged kicking and screaming into the information age. We are being forced to apply some common sense and reassess our entrenched assumptions. It is important we get the last of the four C's in place.

    The most beautiful things on earth, round brilliant cut diamonds, are more a product of their proportions (Cut) than either their Clarity or Colour. After all can you tell the difference between the top six clarity grades IF-SI or the top 3 colours in a setting under normal everyday viewing conditions? I cant, yet I ask my customers trust me and to pay more even if the stone does not look better?



    Leakage is the great enemy, as we all know having grown up with ray trace diagrams. Measuring loss out the pavilion and girdle would seem a simple calculation I can not understand why it was not included this study. And what does the reference to ‘darkness’ mean in this study? Poor make results in a dull white outer region around the upper girdle facets. Even the nail head (45* pavilion angle) does not look dark to a real person. Only a gemmologist examining a gem with a dark lens just in front of the stone, sees a dark table area.

    Light leaks out the back of a diamond in varying degrees as a response to how poor the proportions of the crown and pavilion are in relation to each other. The upper girdle area is the first to suffer, then the region just inside the table*outside the table reflection. Thirdly the kite facets leak, and finally if the stone is a very bad combination of deep pavilion and steep crown angles the star facets may also leak. Measuring this leakage would have enhanced the WLR and helped many people gain a greater understanding of this complex subject.

    Evenly weighted source lighting

    The GIA chose a diffuse hemisphere of even white light directed on thecrown of it’s virtual diamonds.
    If light was supplied in the same way WLR has been measured, then I suspect the results would be very different. Consider this challenge to the assumptions about lighting:- .

    An estimated 70 per cent by value of all diamonds are worn in rings. Rings are generally lower than eyes and are viewed from above (supporting the WLR assessment method). Lighting - natural and artificial - tends to come from above, not evenly weighted. I suspect if lighting was provided somewhat more the way WLR is measured, the WLR or brilliance results would favour traditional 'ideal cuts' with big tables and deeper stones with extremely shallow crowns and small tables.

    It would be interesting if the GIA could test some of these concepts. I suspect the unusual peaks in figure 7 will be quite different simply by providing the type of lighting that most likely will be chosen for the ‘fire’ and scintillation study.

    In case you have never noticed, ‘fire’ does not exist in a floodlit environment. Actually it is there, but you just can not see the dispersed colours because they are washed out by the abundance of white light emerging from every possible direction. This is a marketing problem when photographing diamond jewellery. Fire is impossible to capture with the flood lighting commonly used.

    I have suggested the GIA provide an additional equal amount of spot light source from directly above, 15* and 30* away from perpendicular for all three studies, brilliance, dispersion and scintillation.

    The simple profile images from Bruton’s “Diamonds” depict light emerging from shallow, ‘ideal’ and deep diamonds. The shallow profile throws the light to the sides; the ideal evenly disperses light, while the deep stone throws light straight back up. Diffuse lighting favours shallow crowns because these stones will direct light like a reflecting telescope mirror, straight up. Whereas deeper and ideal-cut stones will perform better if light supply is weighted more from above than the GIA’s evenly provided lighting. I hope this helps simplify my point.

    Perhaps the anomalous 23* crown angle from Fig. 7 performs so well for the following reason. A diamond with a 23* crown angle directs light in a more perpendicular direction than a 34.5* crown. The weighting for returned light favours this angle. The 23* crown will ‘face-up’ brighter, but will preform unsatisfactorily from other than the face up direction. I have done some simple 2D calculations, which appear to provide a confirmation.

    This, however, is an esoteric discussion as a 23* crown stone has very little dispersion (fire) and less interplay of light between crown and pavilion facets (scintillation). Its acute girdle angle will mean such stones will be prone to chipping.

    Grahame can we lift out the line sketches from Bruton ???????


    Neglecting the positive effect of reflections appears invalid. Any reflection above the crown is still a positive return of light. Light reflected from crown facets upwards will contribute more to WLR than light reflected from the table. Whatever the effect and impact, light reflections from diamond contribute to brilliance. The assumption that removing the impact of 'glare' because “the relative variation was essentially unchanged” is not a valid argument in a study as rigorous as this.

    In the future work on fire and scintillation the GIA must include reflections. A ray of dispersed light will not be visible if its respective colour is masked by a brighter reflection of ‘glare’.

    It seems appropriate to use the same lighting model for all three studies.


    The flexibility of a range of ideal proportions will help cutters maximise both yields and beauty. With a meaningful and universally accepted grading system, the market will place a value on each grade. Cutters will be rewarded for producing good looking diamonds. Consumers will be offered better-looking diamonds in which they will have more confidence.

    The single crown and pavilion angle target provided by Tolkowsky is so wasteful that cutters hardly ever achieve it* less than 3 per cent of a sample of 67,621 (actually, only 76 out of 67, 621, or less than 0.12 percent). My linear target, supported by the GIA research, is far easier for cutters to achieve. Such information is easily programmed into Sarin* and other smart cutting aids. The result of cutters adding more value will raise the over all standard of beauty for all round brilliant cut diamonds. I suspect added value will be genuinely created.

    Marketplaces seek standards and give those standards values based on supply and demand. Introducing a new standard will open up price differentials*more for the best and less for the worst. This has been a missing ingredient in the diamond market. Martin Rapaport has provided us with a service to quantify the first 3 C’s of Clarity, Carat weight and Colour. For decades Cut has been a haggling point between buyers and sellers. The problem is that very few people can actually pick a well-cut diamond. This may be a contentious statement, however the evidence of conflicting standards of the many existing cut grading systems is somewhat damming. If there is any further doubt, then examine the appalling variances in the published works. (Table 2 P163)

    Given a number of effective combinations of cut ‘grades’; it would seem a shame to prescribe a hierarchy. This would be especially true if the maximum brilliance, ‘fire’ and scintillation do not all occur from the same set of proportions, as I expect they will not.

    Systematic diamond grading systems do not set prices or values. The effect of supply and demand - the market - sets values. The market will quickly determine which are the most favoured parameters when a new acceptable grading system is introduced.

    Who knows, spready diamonds with high WLR’s may become more valuable than the lower yielding traditional Tolkowsky ‘ideals’? Given the more efficient use of rough for the spread cut, we would have a happy balance: more supply and more demand. The point is no one needs to decide what is the best grade. Our role as experts is to define parameters for the most effective returns of the physical properties of light. To define beauty in such a complex area is to be prescriptive. Different people will have different preferences. We do not need a linear 1-10 type of scale.

    The reason for pointing this out is as the market ‘settles down’ Martin Rapaport and others will assess and report the various values. Grading standards should convey the information in such a way as to allow the market freedom to choose the price it pays for value perceived. There will even be regional discrepancies, just as the East Asian market prefers high clarity with less concern about colour. Geographic and cultural cut preferences also will arise.

    Sales staff with very little understanding of cut will be able to explain the 4 C's with newly designed De Beers sales aids based around a global standard. Consumers will understand more about what they are being offered, and will make better-informed selections. Removing this major area of doubt remains the biggest opportunity for the trade.


    After attending the GIA Symposium in San Diego in June, I understand the difficult position the GIA is in. They have never established a cut grading system (they only report table and total depth %, girdle, culet, symmetry and polish on certificates.) The stance they have taken reflects their venerable position of responsibility. They must get it right the first time!

    I have presented my discovery to the four authors whose paper is the subject of this critique. I have volunteered my assistance to both contribute and to test my theories with the GIA’s accumulated data.
    In a further effort to find a solution to this complex issue you are invited to contribute to an open website discussion at ‘’ (please no commercial activity). You are welcome to collaborate and help devise an unassailable set of standards for Cut.

    [This critique has been published in response to the importance of the original research, and the pertinence of Garry Holloway’s thoughts and arguments. Ed.]

    Address for correspondence:
    Garry Holloway
    Precious Metals
    112-114 Canterbury Rd
    Canterbury Vic. 3126
  5. strmrdr

    Nov 1, 2003
    by strmrdr » May 23, 2005
    Interesting Garry.

    The thing in question here is if the current cut grading is primarly based on the observation study that GIA conducted.
    It is my understanding that it is.
    Is it or isnt it?
  6. Garry H (Cut Nut)

    Aug 15, 2000
    by Garry H (Cut Nut) » May 23, 2005
    Yep Storm, that''s how I read it too.
    But GIA spent a lot of money on their ray tracing and maybe it would be difficult to admit that more than 10 years study was a waste?
  7. JC

    Sep 11, 2003
    by JC » May 24, 2005

    I think that you just set the Pricescope record for the longest post. LOL

Thread Status:
Not open for further replies. Please create a new topic or request for this thread to be opened.

Share This Page