Appellant filed an application serial No. 112,160 on May 15, 1961 for an improvement in the process of spinning hollow filament viscose yarns, which application was a continuation-in-part of an application serial No. 834,673 filed August 19, 1959. The board sustained the examiner's final rejection of all the claims in the case, process claims 1-7, as obvious variations of a combination of two references. That adverse decision is the subject of this appeal.

As appellant discloses, it is well known in the art to produce hollow rayon filaments by extruding a viscose1 solution containing an alkali metal carbonate into an acid bath, wherein the carbonate decomposes to form carbon dioxide. The carbon dioxide inflates the spun viscose stream to make a hollow filament.2 As shown by appellant's example 1, when carbonate-type viscose is extruded through a spinneret having round orifices into a particular spinning bath, a fabric woven and dyed from the yarn thus produced was judged to exhibit "commercially unacceptable" uniformity of dyeing, numerous dark flashes occurring throughout the fabric.

In appellant's process such uneven dyeing is reduced by extrusion of the carbonate-type viscose through slot-shaped spinneret orifices, in which the ratio of length to width is greater than about 5. Appellant states that the more even dyeing obtained:

"* * * is apparently due to the fact that the cross-sections of filament are more uniform, i. e. filaments with abnormal cross-sections are substantially eliminated. * *"

"1. In a process for the formation of a hollow filament of viscose rayon by extruding viscose containing an alkali metal carbonate through a spinneret into a sulfuric acid coagulating and regenerating bath, the improvement which comprises extruding the viscose through the spinneret to form a viscose stream having a slot-shaped cross section at the exit of the spinneret orifice, the said cross section having a width of from about 0.0025 to about 0.005 inch with the ratio of the length to width being above about 5."

The remaining claims more specifically define the length to width ratio of the rectangular orifices and the composition of the viscose. Appellant does not predicate patentability on such additional limitations.

  Picard         1,831,030     Nov. 10, 1931
  Brumberger     1,964,659     June 26, 1934

Picard was cited by appellant in his specification as an example of the carbonate-type viscose used to make hollow filaments. Picard states his hollow filaments are "endowed with a higher covering power,"3 but does not disclose the shape of the orifice used in his spinneret. Brumberger shows a spinneret with rectangular or slot-shaped orifices having length to width ratios greater than 5. Brumberger states his spinneret is useful in:

"* * * all processes for the manufacture of artificial multi-filament yarns, such as the cupra-ammonium, viscose, cellulose-acetate, cellulosenitrate, and any other processes where cellulose and its derivatives are used to produce multi-filament yarns."

The examples specifically described in Brumberger use ordinary viscose which does not contain a carbonate, and accordingly would produce a flattened solid viscose rayon filament. Brumberger teaches that by the use of rectangular orifices, cellulosic yarns having greater covering power will be produced, and that the harshness or softness of the yarns can be controlled by variations in the size or shape of the orifices.

The examiner rejected the claims as unpatentable over Picard in view of Brumberger, seeing no "invention" in carrying out the Picard process using the rectangular orifices of Brumberger. The board considered the rejection as one of obviousness, stating:

"It is obvious to one skilled in the art who wanted the greater covering power of the filaments obtained by Picard and either the additional covering power of Brumberger's ribbon-like filaments or their greater harshness or softness, depending on the denier, that such results could be obtained by using the spinnerets of Brumberger in the spinning process of Picard.

* * * * * *

"We think it is obvious that Picard's filaments could be made in ribbon-like form as taught by Brumberger and the desirability of doing this is also obvious."

Appellant argues that it was to be expected that the use of the Brumberger spinneret in the Picard process would give yarn having very poor dyeing properties since each process was known to produce yarn which dyes non-uniformly. Appellant's only support for the argument lies in the following statements in his specification:

"* * * This result [more uniform dyeing] is surprising since, as is well known to those skilled in the art, the substitution of slot-shaped orifices for round orifices in conventional viscose spinning leads to difficulty in controlling uniformity of dyeing due to the fact that the shape of the filament cross-section tends to vary more with slot-shaped orifices than with round orifices."

In effect, the improvement in uniformity of dyeing is to be looked on by us as an unexpected result. In response to the board, appellant argues that the art does not disclose what effect the use of the Brumberger spinneret in the Picard process would have on the covering power and harshness or softness of hollow filament yarns. In our view, that response carries little weight since the rejection is not for lack of novelty under section 102, but for obviousness under section 103. "Obviousness does not require absolute predictability." In re Moreton, 288 F.2d 940, 943, 48 CCPA 928, 933.

While appellant states that the prior art knew the filament cross-section of ordinary viscose spun through a rectangular orifice would tend to vary, appellant's specification does not aid us by showing an example of such filaments, nor has any affidavit been submitted to clarify that property of the Brumbergertype flattened filament. As to the various types of cellulosic filaments produced by the use of rectangular spinneret orifices, Brumberger states that they have "a cross-section of very long and thin shape," "generally a rectangle bent somewhat or folded over on itself," and "of the same general shape and relative proportions as the orifices producing the filaments." Brumberger contrasts such shape to the irregularities in ordinary viscose spun through round orifices:

"* * * circular orifices * * * [form] filaments having a cross section usually round, but sometimes a cross section may have been of horseshoe shape or kidney shape, or the like."

The solicitor contends that the reasonable inference of that disclosure of Brumberger is that the flattened solid viscose filaments have a uniform cross-section, and thus more uniform dyeing would be expected; consequently one wishing to produce better dyeing in hollow filaments would use the rectangular orifices of Brumberger.

Except for that inference, the Patent Office has not challenged appellant's statement of what the prior art knew of the irregularities in a flattened solid viscose filament. Indeed, the board assumed arguendo the correctness of appellant's statement of irregularity in such filaments, but did not find it conclusive of non-obviousness. Thus that assertion in appellant's specification that flattened solid ordinary viscose filaments "tend to vary" in cross-section must be considered in our determination of obviousness or non-obviousness under section 103.

While a selection of certain facts in this case tend to a conclusion of non-obviousness and others taken alone may show obviousness, the conclusion required under section 103 must be grounded on a weighing of all the facts.

Upon review of the weight accorded all the evidence below, we do not think the board erred. We think there is adequate reason to conclude that the claimed process would be obvious to one of ordinary skill in this art in view of the teachings of the references. While merely for the purpose of obtaining uniformity of dyeing, the process may appear to be non-obvious, such a view does not accord weight to all the facts. Obviousness is not to be determined on the basis of purpose alone. As against appellant's "showing" that slot-shaped orifices result in flattened solid ordinary viscose filaments the cross-sections of which "tend to vary," we must consider Brumberger's teaching that his slot-shaped orifices are useful for "all processes for the manufacture of artificial multi-filament yarns," and that the shape of his filaments conforms to that of the rectangular orifices used, in contrast to the shape produced by round orifices. Further, one in this art could equally choose to obtain the property of greater covering power for the Picard carbonate-type viscose filaments by using rectangular-shaped orifices as taught by Brumberger. This is particularly the case since Picard, like Brumberger, is interested in the production of filaments "endowed with a higher covering power, * * *." As the board stated:

"* * * undyed filaments are still a useful product or the filaments could be colored, as they often are, by putting a dye or pigment in the spinning solution. As described by appellant, the dyeing problem is one of after-dyeing spun filaments."

The process as claimed does not admit of a purpose which would make it unobvious per se, and the degree of improved uniformity in dyeing in subsequent processing, "apparently due to the fact that the cross-sections of filament are more uniform," is not sufficient to tip the balance in appellant's favor. Nor on such facts are we in doubt such that it should be resolved in favor of appellant. For the foregoing reasons we affirm the decision of the board.

Affirmed.

RICH, J., concurs in the result.