Opinion
Patent Appeal No. 9108.
August 1, 1974. Rehearing Denied October 17, 1974.
Thomas M. Ferrill, Jr., Allen V. Hazeltine, Blue Bell, Pa., attorneys of record, for appellants; Sheldon Kapustin, of counsel.
Richard C. Sughrue, J. Frank Osha, Darryl Mexic, Washington, D.C. (Sughrue, Rothwell, Mion, Zinn Macpeak), Washington, D.C., attorneys of record, for appellees. Martin G. Reiffin, Alvin J. Riddles, of counsel.
Appeal from the Board of Patent Interferences.
Before MARKEY, Chief Judge, and RICH, BALDWIN, LANE and MILLER, Judges.
This is an appeal from the decision of the Board of Patent Interferences awarding priority on all counts, counts 2 through 17, in Interference No. 93,544 to Williams et al. (Williams), the senior party. The interference involves Williams' patent No. 3,012,725, for "Electronic Digital Computing Devices," granted December 12, 1961, on an application filed July 22, 1957, as a division of an application filed June 1, 1950, and an application of Eckert et al. (Eckert) for a "Binary Automatic Computer" (acronym: BINAC), serial No. 179,782, filed August 16, 1950. We dismiss the appeal as to count 16 and affirm on the other counts.
The questions before the board and still in issue here are: (1) Eckert's right to make counts 2-15 and 17, (2) Williams' right to the benefit of the filing date of an earlier-filed British provisional application, and (3) Eckert's alleged conception and diligence, and actual reduction to practice. On the first issue, the board ruled that the Eckert application did not support counts 3-7, 9-12, 14 and 17, but that it did support counts 2, 8, 13 and 15. It denied Williams' claim to the benefit of the earlier British provisional application. The board additionally ruled that Eckert did not have an actual reduction to practice prior to Williams' effective filing date of June 1, 1950.
In his brief, Eckert stated that he does not now contest the board's holding that he is not entitled to make count 16.
The invention involves an improvement in an electronic digital computer wherein instruction words may be stored in pairs in a single location in an addressable memory. As the Williams patent indicates, that feature permits use of a reduced number of memory locations and a saving in cost because instruction words in a program frequently involve less than half of the bit positions normally available in each location. The systems of the Williams patent and the Eckert application differ in a manner which will be further explained below. An adequate understanding of the invention may be obtained from inspection of one of the more comprehensive counts, count 11, which reads:
11. An electronic binary digital computing machine including
[a] at least one data word storage device having a plurality of discrete address storage locations each of a digit capacity equal to the digit length of one machine number word,
[b] signal-controlled address selection means for rendering any required one of said address storage locations accessible for reading-out or writing-in thereto under the control of an applied address selection signal defining said required address location,
[c] signal-controlled selection means for restricting the data selected by said address election means to a chosen one of a plurality of integral fractional parts thereof,
[d] a control system for governing the operation of said machine during each computing cycle in accordance with the form of an applied instruction word signal, said instruction signals being each of a digit length equal to one of said integral fractional parts,
[e] said control system including means for selecting a series of instruction word signals in turn from said data word storage device each under the control of a binary coded instruction address selection signal, said binary coded instruction address selection signal including at least one lowest significant digit providing a control signal for said selections means and at least one further more significant binary digit for providing an address selection control signal to said signal-controlled address selection means. [paragraphing and emphasis added]
The first issue considered by the board was Eckert's right to make the counts. On that matter, the board stated:
Williams' principal contention is that although the disclosed Eckert apparatus stores two instruction words at a single address, the one occupying the most significant digits portion is always read out and used first and then the one occupying the least significant digits portion is utilized so that there is no selectivity or choice in the order in which the words are read out and utilized. There seems to be no contention that this is an incorrect interpretation of the Eckert disclosure. Williams urges that all of the counts require selection or choice by virtue of which either of the two words could be read out and used first. Eckert urges, and the Primary Examiner held, that the counts are not limited to require this option.
It then discussed the specific language in each count which Williams urged was unsupported. The board concluded that Eckert did not support counts 3-7, 9-12, 14 and 17, but that counts 2, 8, 13 and 15 were so broad that Eckert could support them.
Before us, Eckert challenges the decision against him as to counts 3-7, 9-12, 14 and 17. While he obviously accepts the board's favorable decision as to counts 2, 8, 13 and 15, he does not offer any specific supporting argument. Eckert does not question the board's understanding of the operation of his computer. Williams urges in detail, as he did before the primary examiner and then the board, that Eckert has no right to make any of the counts.
Turning to the parties' disclosures, both Eckert and Williams provide means whereby a pair of instructions, including a left half instruction and a right half instruction, are stored in a single location having a digit capacity of one data word. The Williams computer includes apparatus which permits selective execution of either the left half instruction disposed in the left half of the location or the other instruction disposed in the right half of the location. The non-selected instruction may be either excluded from the program or executed after the selected instruction. In the Eckert BINAC computer, the multiple instructions in the single location are executed in a fixed order during the machine cycle.
In the Williams computer, after the instruction pair is extracted from memory and entered into the instruction pair holding register, the next step is entirely dependent upon a control bit in the control store. If the control bit has a first value, the left half instruction is gated from the instruction pair holding register to the L and F staticisor (instruction register) to be executed. If the control bit has a second value, the right half instruction is gated and executed instead. The half of the instruction pair which is not gated out in response to the control bit is not acted upon by any apparatus in the Williams computer. If it is desired subsequently to execute the "other" instruction in the pair of instructions, the Williams computer must go back to the memory again, re-extract the same instruction pair and repeat the selection process. This arrangement provides added flexibility by permitting the use of an instruction half in any sequence relative to its companion half or any instruction half in any other memory location.
Eckert's disclosed computer permits no such flexibility. Once a pair of instructions is entered into the instruction holding register or location, both halves of the pair are invariably gated out and executed in the inflexible sequence of the half in the most significant digits region followed by the half in the least significant digits region.
The board found that counts 3-7, 9-12, 14, and 17 all required that the apparatus offer a selection or choice between the halves or parts of the instruction words of a pair. As an example, it found that such a choice was required by the structure defined in the following expression in count 3:
* * * subsidiary signal-controlled selection means for restricting the data selected by said address instruction means to a chosen one of a plurality of discrete equal fractional parts of said selected data. [board's emphasis]
Particular emphasis was placed on the term "chosen one" as requiring "the ability to provide access restricted to any one of the stated plurality of discrete parts at a given time." The board then held that the disclosed Eckert apparatus could not satisfy this requirement since Eckert did not disclose a way in which a choice between the two halves of a particular instruction pair could be accomplished.
As a further example, the board referred to the following terminology in count 14:
* * * means under control of said other selecting manifestation for selecting said one portion or said other portion.
Here the board found that a selection between one portion or the other is required, and the board found that the disclosed Eckert computer was not capable of making such a selection.
The board also regarded the requirement for "choosing" or "selecting" a particular half of the word pair to be beyond the teaching of Eckert in connection with the other counts which it found he could not make. We agree with the reasoning of the board regarding counts 3-7, 9-12, 14 and 17 and find no reversible error in its ruling that Eckert cannot make them.
The board held that Eckert had the right to make counts 2, 8, 13, and 15. Williams contends that the board erred in this respect. Williams states in his brief:
Each count expresses the important distinctions discussed previously by the use of such ordinary words as "choose", "select" and "restrict". Given their ordinary meaning, the words "choose" and "select" connote the ability to pick among alternatives, a feature present in the Williams computer but absent in the BINAC computer. The word "restrict" simply connotes taking less than the whole — also a feature present in the Williams computer but lacking in the BINAC computer. Because each of the counts contain[s] limitations based on these terms, it is submitted that the Eckert et al. disclosure supports none of the counts. * * * [emphasis in original]
He further argues specifically why he thinks Eckert does not support counts 2, 8, 13 and 15. Eckert does not provide us with a discussion of the terminology in these counts which the board held were supported, nor does Eckert respond to the detailed arguments which Williams makes concerning these counts.
Count 2 reads:
2. An electronic digital computing machine including
[a] at least one data word storage device having a plurality of discrete storage address locations each of a digit capacity equal to one machine data word,
[b] signal-controlled address selection means for rendering any required one of said storage address locations accessible for reading-out therefrom or writing-in thereto under the control of an applied address selection signal defining said required address location and
[c] subsidiary signal-controlled selection means for restricting the data selected by said address selection means to a fractional part only of said selected data. [paragraphing and emphasis added]
The similarity to count 3, above, is apparent. In finding count 2 supported by Eckert, the board distinguished it from count 3 because "[n]o selection or choice between plural portions is required." Williams argues that the board erred because Eckert uses the multiple portions in the selected data word position in sequence, and is incapable of restricting the use of the selected data to only a left or right half thereof. That analysis agrees with our own understanding of the count and of the mode of operation of the disclosed Eckert machine. Hence, we agree with Williams that Eckert cannot make count 2.
Count 8 reads:
8. An electronic digital computing machine arranged for operation with serial mode number word signals and including
[a] at least one data word storage device having a plurality of discrete word storage locations each of a digit capacity equal to one machine number word length,
[b] rhythm control means for controlling the machine operation as a series of sequential minor cycles or beats each of a time duration suitable for accommodating one machine number word signal,
[c] a signal-controlled address selection means for rendering any required one of said storage address locations accessible for reading-out therefrom or writing-in thereto during the time of one minor cycle or beat under the control of an applied address selection signal defining said required address location and
[d] subsidiary signal-controlled selection means for restricting the data selected to a fractional part only. [paragraphing and emphasis added]
The board relied on the same reasoning here as with respect to count 2. We again disagree with the board for the reasons set forth above.
Count 13 reads:
13. In a data processing system,
[a] means for storing an instruction manifestation comprising one instruction portion and another instruction portion and
[b] means for extracting from said storing means said one instruction portion or said other instruction portion. [paragraphing and emphasis added]
The challenged portion of the count is paragraph [b]. The board held that this count was so broad that it did not require a selection or choice to be made at any particular time, and that Eckert could make it since the respective portions are extracted separately, albeit always in the same order. Williams states that Eckert's control register which stores the left half and right half instructions corresponds to the claimed means for storing and that "the apparatus which extracts the instructions from the control register is only capable of operating to extract the first portion and the second portion in sequence." Williams further points out, "[i]t does not extract one portion or the other portion." We agree. Count 13 clearly requires a means for extracting, as desired, one instruction or the other instruction portion. The board erred in holding that Eckert could make count 13.
Count 15 reads:
15. An electronic digital computing machine arranged for operation in the serial mode and comprising
[a] a digital storage device having a plurality of discrete address or storage locations each capable of registering a separate item of numerical or coded instruction data of predetermined maximum number of digits in magnitude,
[b] address selecting means for rendering any one of said address or storage locations accessible and
[c] selective read out means for selecting a chosen part only of the data selected by said address selecting means in said storage device. [paragraphing and emphasis added]
We think that "selective read out means for selecting a chosen part only" requires the ability to make a selection or choice of one part only of the data in "one of said address or storage locations" that has been rendered accessible by the "address selecting means." Thus, the limitation in paragraph [c] is not supported by the disclosed Eckert apparatus where both parts of the data in the selected address or location are invariably extracted in the same sequence. The board erred in holding that Eckert could make count 15.
Eckett argues that the presence in the Williams patent of a claim (23) with a more specific recitation of the selection feature than the counts indicates that the latter were intended to claim the invention more broadly, citing Oldroyd v. Morgan, 57 F.2d 358, 19 CCPA 1111 (1932) and Marshall v. Ledwinka, 67 F.2d 495, 21 CCPA 728 (1933). The proposition Eckert finds in these cases appears to be that a patent claim lacking a recitation set out in a second claim is to be interpreted as not including the same limitation. While that proposition cannot be denied, the cases obviously are not applicable here. None of the present counts reads the same as claim 23 with the selection feature omitted. Rather, as is apparent from the above, they define that feature in different language than claim 23, which language Eckert likewise fails to support.
Since we have decided that the board was correct in holding that the Eckert application does not support counts 3-7, 9-12, 14 and 17, and that the board erred in holding that the Eckert application does support counts 2, 8, 13 and 15, the decision awarding priority to Williams must be sustained. We do not reach the issue of Williams' right to the benefit of the filing date of an earlier-filed British provisional application, or the questions involving Eckert's alleged conception and diligence, and actual reduction to practice. Accordingly, the appeal is dismissed as to count 16, and the decision of the board awarding priority to Williams as to counts 2-15 and 17 is affirmed.
Affirmed.