Hackh's Chemical Dictionary, 3d Ed., states under "chemicals": The grade or the purity of chemicals on the market are [sic]:

The board predicated its decision largely on the notion, for which there is no support in the record other than its opinion, that

One skilled in this art would not be expected to employ an impure reactant, but would follow normal chemical practice and use a pure phthalic anhydride in the esterification. Since the references are silent as to any content of ferric ion it cannot be presumed that such ion is necessarily present in the starting material. [Emphasis mine.]

It is very significant that "the references are silent" on the key fact in this case and I shall not labor the point except to reverse the board's observation to say that in view of the silence it cannot be presumed ferric ion is necessarily absent in "pure" starting material.

The board's statement about "normal chemical practice" is absolutely meaningless to me in the context of this case by reason of the vagueness of the term "pure." As the above quotation from Hackh amply demonstrates, "pure" does not normally mean chemically pure for when the latter is intended it is specified. "Pure" in the usual sense is a relative term meaning, according to Hackh, "sufficient purity for general work" and is quite broad enough to encompass as much as 8 ppm. of some impurity. I find further support for this, even with respect to C. P. chemicals, in Chemicals of Commerce, Snell and Snell (Van Nostrand, 1939), p. 1:

The term C. P. stands for chemically pure. Materials bearing this label may be supposed to contain a minimum of impurities, but the term is somewhat general and does not mean specifically 100 per cent pure. Strictly speaking, no chemical is absolutely pure although the known impurities may be expressed as present to the extent of 0.0001 percent. ***

The term technical or technical grade is applied to products as they are commonly produced on a large scale. The degree of purity of technical products varies with the individual substances and depends on the ease with which contaminants are removed during the manufacturing process. Such relative terms as crude and refined are also used.

Appellants, presumably skilled in this art and using chemical terminology knowledgeably, made the following statement in their specification, apparently so far overlooked by everyone:

The brominated product [TBPPA] is obtained contaminated with an impurity, either physically absorbed or molecularly completed, such that by the usual 307-345-68—7

methods of purification the contaminant is not entirely removed. [Emphasis mine.]

It seems to me we are justified in assuming that the "pure" chemical the board expected to be used in the art would be such a one as has been subjected to the "usual methods of purification" and we are then left with the very problem, unrecognized in the art but nevertheless existing, which was discovered and solved by appellants. It is my opinion from study of the prior art references that it was a very unobvious problem or source of difficulty.

*

I am not able to accept Judge Almond's attempted close reasoning in support of the majority for the further reason that it is predicated, at least in part, on a non-existent supposed "concession" by appellants that TBPAA had previously been used to form polyesters, presumably with success. The implication is that it was conceded by appellants to have been successfully used, yet we find appellants quoting from KirkOthmer's Encyclopedia to the effect that "bromo-derivatives of phthalic anhydride [for fire retarding] ** have been either unsuccessful or economically unattractive." But aside from that reference to Kirk-Othmer, what the specification actually says is not that TBPAA has been used but that it has been "disclosed *** for use" or "proposed," which is a very different thing. Such disclosures and proposals put TBPAA legally in the prior art but this is not a real use from which, as Judge Almond attempts to do, one can deduce "that the prior art had [commercially] available TBPAA having relatively low ferric iron content." When this deduction loses its foundation, the argument predicated on it collapses.

Equally unconvincing is the majority opinion's attempt to reason, from the disclosures in Phillips and Wismer that TBPAA can be used, that the Kirk-Othmer statement is unpersuasive, at least to anyone familiar with the way disclosures in patent specifications like those relied on come into being. It is quite true those two patents ignore the existence of any problem just as though none existed. And why not, if it was not appreciated that there was one? It is easy to propose use of a chemical without knowing how it will behave. Both Phillips and Wismer merely list TBPAA among numerous possible polycarboxylic acid anhydrides which might be used. Neither indicates it was ever tried. Anyone skilled in this art could make up such a list. Such haystack disclosures are without evidentiary value. One cannot deduce anything from them, as from a commercial reality, and that is what the majority opinion is attempting.

On the other hand we have the experimental data of appellants' Example V showing the actual effects of the ferric iron content. Its essence is as follows:

In a method of polyester preparation similar to that employed in the previous Examples, wherein 1:1:2.2 molar relationships of TBPAA, maleic anhydride and diethylene glycol were utilized, the following six samples of TBPAA were employed:

Ferric Iron Determination-PPM.

(a) TBPAA, obtained commercially from Michigan Chemical Company-- 34 (b) Technical TBPAA, prepared as described in Example III‒‒‒‒‒ 8 (c) TBPAA that had been methanol-conditioned by the method of Example I(a ) ---.

(d) TBPAA that had been methanol-conditioned by the method of Example I(a) --

(e) TBPAA conditioned by the sodation method of Example IV (a).

(f) TBPAA conditioned by the xylene crystallization method of Example III (a)

5

21

1.2

4

No polyester could be produced with the use of the Sample (a) above TBPAA component. The use of Sample (b) did not consistently produce polesters. However, Samples (c) through (f) consistently produced polyesters as previously demonstrated in Examples I-IV, above.

The references to conditioning of the TBPAA are to special methods of purification to remove residual ferric iron disclosed in the application at bar and in other applications.

The majority's attempt to shrug off this evidence, particularly the results obtained with Example V (b), 8 ppm. impurity, as being "equivocal" and so not showing the existence of a problem impresses me as a tour de force. Anyone running a chemical plant sees nothing "equivocal" in inability to get consistent results with a given raw material or reactant. It is such a problem as can practically lead to a plant shut-down and abandonment of a process altogether. Besides, that is exactly what the Kirk-Othmer publication suggested, notwithstanding the bland assumption to the contrary of patentees who never tried it.

In summary, the only question we have to decide is the obviousness of appellants' claimed discovery of the importance of keeping the ferric impurity in TBPAA, used to impart fire-retardant properties to polyesters, below 5 ppm. That question must be decided on the basis of what the prior art of record teaches. It cannot be disputed, it is not in fact disputed by the majority, nor was it disputed by the board, that the prior art references teach nothing on that subject. The majority view rests only on a complicated rationalization based on hindsight which attempts to show that the skill of the art would make it obvious but, for the reasons above discussed, the underpinnings of that rationalization are inadequate logically to sustain it.

I would reverse.

SMITH, Judge, dissenting.

The invention with which we are here concerned is a detailed, limited improvement in the highly technical art of polymerizing certain designated materials for producing fire-retardant compositions. As stated in the specification:

*** More particularly, it relates to polymerizable compositions comprising polyesters of polymerizable compositions comprising polyesters of polyhyric alcohols and polycarboxylic acid anhydrides, at least a portion of which is tetrabromophthalic acid anhydride [TBPAA] that gives the products produced therefrom said fire-retardant characteristics; and to processes for the preparation of such polymerizable compositions.

The specification also provides the further background information that:

The production of flame-retardant polyester resins is of great commercial importance. The expanded use of polyester resins in the coating composition field, which includes increasing application to wall members, structural panels, pipes, electrical contacts and the like, requires that the coatings be resistant to fire and heat deterioration.

Specifically, appellants do not assert nor claim anything but a very specific improvement relating to the critical limits of 5 ppm of the ferric iron impurity in the TBPAA which they assert permits the incorporation of TBPAA in a polyester. The specification characterizes the resultant polyester resins as having "remarkable flame-retardancy when cured." Appellants' position is supported in the specification by detailed and comparative examples of the reactions involved and by test data set forth therein as resulting from controlled experiments which in turn are fully described.

The record herein seems to me to support appellants fully in the following statement in their brief:

Appellants admit that the art (as represented by the Lundberg, Phillips et al, and Wismer et al references) not only knew that tetrabromophthalic acid anhydride (TBPAA) could be used to prepare polyesters, but was aware that TBPAA could be used in the preparation of polyesters with fire-retardant properties.

A practical industrial problem existed, however, in the fact that there was difficulty in preparing polyesters from commercially available TBPAA.

Appellants, thereupon, directed their efforts to solving this industrial problem and made a triple-faceted discovery:

1. The cause of the difficulty in preparing polyesters with TBPAA resides in the presence of ferric iron. The following statement made by this Court in In re Antonson, 47 CCPA 741, 124 U.S.P.Q. 132, 133, is apropos: "*** the inventive act which entitles an applicant to a patent resides as well in the discovery of the source of trouble as in the application of the remedy." citing Eibel Process Co. v. Minnesota & Ontario Paper Co., 261 U.S. 45, 67.

2. The remedy resides in utilizing TBPAA with no more than 5 ppm (parts per million) of ferric iron associated therewith. The criticality of 5 parts per million is illustrated in Example V of the specification *

3. An unexpected advantage of using TBPAA associated with no more than 5 parts per million of ferric iron is that fire-retardant properties far in excess of expectation are achieved. Table I * * * clearly shows the unexpected advantage with respect to fire-retardant properties achieved by the use of TBPAA associated with a limited amount of ferric iron in relation to the use of TC1PAA (tetrachlorophthalic acid anhydride), PAA (Phthalic acid anhydride), and HCINAA referred to by the Examiner as "chlorendic acid". Note particularly the column "% Consumed". *

With the foregoing as a background, the issue which the majority resolves against appellants is that of obviousness of the claimed invention under 35 USC 103.

The examiner's analysis of appellants' invention and the prior art was adopted by the board in affirming the rejection and this in turn underlies the majority's affirmance of the board's decision. It seems to me, however, that the examiner's analysis substitutes his view, by a process of hindsight reconstruction, for what Congress has provided in 35 USC 103 as the standard by which to determine what would have been obvious to one of ordinary skill in this art at the time of appellants' invention.

Appellants readily admit that of the art cited by the examiner, the Lundberg, Phillips, and Wismer patents are representative of the prior art on which appellants' improvement is based. Appellants point out, however, that in addition to not recognizing the esterification problem faced by industry or its solution, not one of these references recognizes the special significance of limiting the iron impurity of TBPAA. Lundberg does not mention TBPAA and Phillips and Wismer merely disclose TBPAA as "one among many" of the materials which may be used.

The examiner in his answer ignored the precise limitations in appellants' disclosure and claims and incompletely described appellants' invention as follows:

The invention relates to a process of preparing fire retardant polyester resins from a glycol, a dicarboxylic acid which may be a-ß-ethylenically unsaturated, and at least 10 mol per cent of tetrabromophthalic anhydride of a specific ferric iron content. The polyesters, when unsaturated, may be combined with unsaturated monomers and cured.

It should be noted that the examiner at this point apparently did not consider what appellants have asserted to be the critical limitation in their discovery, i.e., that producing the product depends on the incorporation in the product of the minimal stated amount of TBPAA having the ferric iron impurity in the TBPAA not above 5 ppm.

Thus, appellants' "invention as a whole," which section 103 requires us to consider in arriving at a determination of the issue of obviousness, has a two-fold aspect: (1) an appreciation of a problem which, on the record, was unknown to the prior art, and (2) its solution.