Examiners occasionally take procedural shortcuts when relying on foreign-language prior art, but those shortcuts can create a record that is vulnerable from the first office action through appeal. One such scenario is where an examiner cites a non-English reference, provides no translation of the text, and instead relies primarily—or exclusively—on the figures. While figures can, in some cases, convey structure sufficient to support a rejection, the governing law does not relax the examiner’s burden to establish a factual basis for the rejection. When the record lacks a translation sufficient to support the examiner’s factual findings, the rejection risks resting on speculation rather than evidence.

The Manual of Patent Examining Procedure recognizes the importance of accurate understanding of foreign references and contemplates the use of translations where necessary to establish what the reference teaches. That principle is not merely procedural formality; it is tied directly to the examiner’s burden to make a prima facie case. The Federal Circuit and its predecessor court have been clear that factual findings must be supported by evidence in the record, not assumptions. “A rejection . . . must rest on a factual basis.” In re Warner, 379 F.2d 1011, 1017 (CCPA 1967). The court continued: “The Patent Office has the initial duty of supplying the factual basis for its rejection. It may not . . . resort to speculation, unfounded assumptions or hindsight reconstruction to supply deficiencies in its factual basis.” Id. When an examiner relies on untranslated foreign-language material, particularly to support nuanced functional limitations, that line is easily crossed.

A recent appeal (Appeal 2026-000304, Application 18/363,246) provides a clean illustration of how this issue develops and how it can ultimately result in reversal. The application at issue involved a butterfly valve assembly including a manual override with an override shaft and locking key. The claimed configuration required three distinct positions—fully open, fully closed, and a neutral position in which relative rotation between components is permitted—and three corresponding recesses that cooperate with the locking key. The structural and functional interplay between these elements was central to the invention.

Claim 1 on appeal:

A butterfly valve assembly comprising:
a butterfly element configured to control fluid flow through a main conduit; and
a manual override that includes an override shaft and a locking key, wherein the override shaft and the locking key are operable to lock the butterfly element in a fully open position or a fully closed position;
wherein the override shaft is rotatable between a first neutral position in which rotation of the butterfly element relative to the override shaft is permitted, and a second override position in which rotation of the butterfly element relative to the override shaft is prevented by interaction of part of the butterfly element with part of the override shaft;
wherein the override shaft is rotatable to a third override position in which rotation of the butterfly element relative to the override shaft is prevented by interaction of part of the butterfly element with part of the override shaft;
wherein the override shaft includes:
a first recess formed on an outer surface thereof that mates with the locking key to lock the override shaft in the first neutral position;
a second recess formed on an outer surface thereof that mates with the locking key to lock the override shaft in the fully open position; and
a third recess formed on an outer surface thereof that mates with the locking key to lock the override shaft in the fully closed position

The examiner rejected the claims under § 103 based on a combination including a Japanese utility model publication in view of other references. The examiner did not provide any translation of the Japanese publication – not even a machine translation. The examiner relied on annotated figures to support assertions that the reference disclosed an override shaft with three positions and corresponding recesses, mapping specific numbered elements in the figures to the claimed features. The examiner asserted that certain pins and recesses in the figures corresponded to a “neutral position” as claimed.

The appeal brief challenged both the substance of the mapping and the procedural deficiency. Substantively, the appellant argued that the Japanese reference disclosed only two positions (open and closed) and lacked the claimed neutral position. Procedurally, the absence of any translation meant that the examiner’s interpretation of the figures, including the meaning of labels and the function of depicted components, was unsupported by the record.

The Board called out the lack of evidence to support the examiner’s assertions. The Board analyzed the claimed requirement of three distinct positions, including a neutral position permitting relative rotation, and found that the examiner’s rejection did not satisfy those limitations. While the Board considered that the configurations identified by the examiner corresponded to only two operative positions and did not disclose the claimed neutral state, the Board’s treatment of the untranslated reference was dispositive. The Board expressly noted that the Japanese reference “is a reference published in Japanese, for which no English-language translation (not even of the Abstract) has been provided in the context of the rejection.” The Board further observed that the rejection relied on figures “even though an English translation of the text and annotations in the figures has not been made of record.” Without that translation, the Board concluded that “it is not evident that these figures, devoid of any facilitating translation, support various of the Examiner’s affirmations.” The Board then refused to fill the gap, stating that it would not “engage in speculation or accept assumptions to cure deficiencies in the Examiner’s fact finding,” and reiterated the Warner standard requiring a factual basis.

This example shows how examiner shortcuts, including skipping a translation and relying on an asserted interpretation of figures, can propagate through prosecution and become dispositive on appeal. The deficiency was not cured by the examiner’s answer, nor by citation to additional references, because the core factual predicate of the reference’s teaching was never properly established. The result was a reversal of all rejections.

For practitioners, the takeaways are both tactical and strategic. At the Office action stage, identify early whether a foreign reference is being used without an adequate translation and to force the issue onto the record. Where the rejection depends on nuanced functional language, the absence of a translation is particularly vulnerable.

In drafting responses, it can be effective to separate the argument into two layers: first, that even accepting the examiner’s interpretation, the reference does not disclose the claimed limitations; and second, that the interpretation itself lacks evidentiary support, where the second layer may be strategically held for appeal. Quoting Warner and similar authority can be useful to frame the examiner’s burden in terms that the Board will later recognize. It is also worth emphasizing discrepancies between what can be unambiguously gleaned from the figures and what requires textual context, thereby highlighting where speculation is occurring. Present the Board with a clear choice: either accept unsupported factual assertions or enforce the requirement that rejections be grounded in evidence. The Board’s willingness to reverse here suggests that, when properly framed, this can be a compelling argument.

From a drafting perspective, this line of cases reinforces the value of clearly articulated functional relationships in the claims and specification. The more precisely the claim defines distinct operational states and interactions, the more difficult it becomes for an examiner to map those limitations onto ambiguous or partially understood prior art, particularly where language barriers exist. Detailed descriptions in the specification can also provide a strong foundation for distinguishing prior art during prosecution.

Ultimately, reliance on untranslated foreign-language references is not inherently improper, but it carries risk for the Office if the record does not support the asserted teachings. That risk can be leveraged by applicants who are attentive to the evidentiary gaps and disciplined in building a record that exposes them. This case shows that what may begin as a minor shortcut can become a decisive reversible error when scrutinized under the proper legal standard.