T 0702/99 (Cosmetic/SHISEIDO COMPANY LIMITED) 03-12-2003

1. The appeal is admissible.

Assessment of test evidence.

2. In cases concerning products such as cosmetics, in which applicants or patentees seek to establish that their inventions have an improved "feel" over the prior art, or opponents seek to deny such an improved "feel", it is common for one or more parties to file evidence of comparative tests conducted by a number of persons. It is naturally essential for such tests to be made under conditions which ensure maximum objectivity on the part of those conducting the tests and who may be required at a later date to give evidence in proceedings.

3. Such tests are often carried out by employees of a party. The use of independent persons would naturally tend to carry more weight but would understandably be more difficult to arrange. While the employment relationship means that the testers cannot be described as independent witnesses, the use of employees may not be objectionable per se as long as the test conditions are designed to ensure that, just as if independent persons were used, the employees are not biased by prior knowledge of either the products under test or of their employer's expectation of the result of their tests. Thus it is always desirable that such tests can be shown to be "blind"; that the testers have had no part in the making of the claimed invention or research leading up to the invention or the patenting procedure; and that the tests have been conducted in the strictest conditions - for example that no- one has given any or all of the testers any advance information, that each tester performs his or her test in the absence of the other testers, and that their opinions are accurately recorded. Since such evidence is opinion evidence and thus inherently subjective, its value lies in the number of similar or same opinions and the tribunal faced with such evidence will naturally be seeking to judge the objective value of a number of subjective opinions.

4. The presentation of such evidence must also be accurate, as the present case illustrates, but the format of the presentation is of secondary importance - a carefully prepared report and/or table may convey as much information as a large number of statements from each tester containing much the same information. However, in cases where a report and/or table is used, a statement (either within the body of the report or in the form of a separate witness statement) from the organiser of the tests detailing the test conditions as well as the results can only assist in assessing the objectivity of the test process and, if it demonstrates sufficient objectivity, it can enhance the value of the test evidence. Parties to proceedings should adopt the same standards in the preparation of such test evidence as they should in the preparation of experimental evidence.

5. The various items of test evidence on file in the present case are as follows.

5.1. The patent in suit describes "an organoleptic test" in which a sample softening lotion (Example 1 prepared according to the claimed invention) and Comparative Examples 1 to 4 (representing prior art lotions) were evaluated by a panel of ten persons, described as experts, and graded according to how many of those ten found a particular composition "good" for moistness or lack of stickiness. Only the Example 1 lotion was graded "good" for both qualities by eight or more of the ten experts.

5.2. A report, filed by the patentee with its letter of 5 July 1996 in the opposition proceedings, of a further such test comparing the Example 1 lotion with a further Comparative Example 5. The methodology was as in A above and the result showed the Example 1 lotion to be better.

5.3. This produced a response from opponent 2 (in the appeal, a party as of right) in the form of a report enclosed with its letter of 17 December 1996, in which it claimed to have repeated the patentee's test in 5.2 and obtained the opposite result. The report stated the test was conducted by ten persons who graded the products quite simply by which composition each tester preferred with "no difference" as a third possibility and this assessment was made in relation not only to moistness and stickiness but also to three other attributes. However, contrary to the text of the report itself, the table it contained actually showed a majority of the testers preferred the Example 1 lotion of the patent and thus confirmed the patentee's evidence. Also, while the table recorded ten opinions for four of the tested attributes, it recorded twelve opinions for "stickiness". The patentee naturally challenged the conclusions in the report of these tests (in its letter of 2 April 1997). Opponent 2 replied (in its letter of 22 April 1997) that the table in the test report contained two typographical errors and filed a corrected report in which the original table remained but to which was added a further almost identical table in which those errors had been corrected.

5.4. The patentee then filed, with its letter of 23 May 1997, two tables (Tables I and II) summarising a "blind test", again by a panel of ten persons and again comparing the Example 1 lotion (called A) with Comparative Example 5 (called B). This time the testers were asked to grade the compositions against each other and ascribe one of three results (A better than B, A and B the same, B better than A). All found A better than B for both moistness and stickiness. Opponent 2, in its letter of 11 June 1997, attacked the improbability of such a unanimous finding. The patentee countered that attack by filing documents described as "sworn statements" from all but one of the test panel members and the test organiser.

6. Both opponents made a number of criticisms of this latest evidence at the first instance oral proceedings which the Opposition Division did not accept (see its decision at pages 6 to 8). Those criticisms were as follows.

6.1. That the statements of the test panel members were contradictory in that, in each such statement, paragraph 4 said the pH of lotion A was adjusted to 6.6 by KOH whereas paragraph 5 said the testers did not know the formulation of the sample lotions they tested. The Opposition Division however accepted that the information about the pH was only given to the testers after they conducted their tests.

6.2. That sworn statements are not used in Japanese proceedings. The Opposition Division considered this was of no consequence as the statements were made for EPO proceedings.

6.3. That the witnesses who signed the statements, which were in English, had insufficient knowledge of English to understand what they were signing, two of them in fact having signed in Japanese script. The Opposition Division accepted the patentee's submission that the text of the statements was read to the witnesses either in English or, if there were language problems, in Japanese so they all understood what they were going to sign.

6.4. That the unanimous 10:0 assessment in favour of lotion A in the patentee's final tests was not realistic. Reference was made to the patentee's earlier tests which showed eight or more out of ten testers favouring the product of the patent and which could thus have included a large majority rather than complete unanimity. The Opposition Division however felt that complete unanimity did not mean the test was not objective and that it was possible that the differences between the tested products could be such that all the testers made the same decision.

6.5. Finally, that the tests were not conducted "lege artis" - the patentee's tests only tested stickiness and moistness whereas opponent 2's tests also tested three other attributes, namely skin penetration, applicability and sensation on the skin generally. The Opposition Division took the view that, since the tests were blind tests, it could not understand how limiting the tests to two properties of the lotions could have influenced the results.

7. While the Board considers those criticisms made at first instance had some force, it would not disagree with the Opposition Division's decision in as much as no one of those criticisms was in itself necessarily fatal.

8. However, the Board has further criticisms, summarised in paragraphs 8.1 to 8.10 below, of these tests which, taken with those referred to above, must lead to a different conclusion.

8.1. That all the witnesses were employees of the patent proprietor is unsurprising, but it does make the facts mentioned below - that they all signed identical statements on the same date without having any part in their preparation - all the more significant. It affects the value which can be placed on the statements.

8.2. The statements of the panellists are all totally identical save as to their names and length of testing experience. That identical wording clearly indicates the individual witnesses played no part in the preparation of their statements, a fact supported by the decision under appeal (see paragraph 3.1.2) which records that the statements were read to the witnesses before they signed them. Absolutely identical evidence from several persons must always be viewed with caution because of the risk of collusion. However, absolutely identical evidence described by all witnesses in virtually identical words is of very little probative value at all since in such cases it is clear that someone has caused the witnesses to give exactly the same opinion or the same account of the events in question. Even two persons would be unlikely to choose exactly the same words to describe identical experiences: that nine persons would do so beggars belief. The fact all the statements in this case were signed on the same date enhances their "production line" nature.

8.3. The statements are all headed "sworn statements" (i.e. affidavits) but in fact they are clearly not sworn but only signed. Apparently (see the Opposition Division decision, paragraph 3.1.2) sworn statements are not used in Japan, so the witnesses would not necessarily know what "sworn" means in the context of legal proceedings (indeed, if or to the extent that they understood English). It therefore appears that the word "sworn" was used, by whoever produced these "production line" statements, purely to add some weight to the statements. However, the actual effect is the opposite.

8.4. The time lag between the date the tests were conducted (3 April 1997) and the date the statements were signed (31 October 1997) also undermines the value of the statements as evidence. A contemporaneous record was apparently made when the tests were conducted, since two tables summarising the tests in question were filed with the patentee's letter of 23. May 1997. It seems highly unlikely that, without that record to remind them, the individual testers would in the October remember a test occupying only a few minutes which they made the previous April. Thus the statements add little if anything by way of probative value to the evidence in the tables.

8.5. There are marked discrepancies between the years of testing experience given in Table I and in the statements themselves. Of the nine testers' statements, only three give the same length of experience as Table I. Four show massive differences (11 years in Table I but 15 in the statement; 1 year in Table I but 9 in the statement; 2 years in Table I but 9. in the statement; 2 years in Table I but 5 in the statement). Two others (in both cases 8 years in Table I but 9 in the statements) could only be explained by the time lag referred to above. It is also significant that one of these two statements is that of the organiser of the tests, Haruo Ogawa.

8.6. The reference in each statement to the pH adjustment, whether or not considered in conjunction with the subsequent denial in each statement of any contemporaneous knowledge of the composition, strikes the objective reader as distinctly curious. Although the Opposition Division accepted this was information given to the witnesses after the tests, there is no evidence to support this. There might be other information about the lotions given to the testers after the event, so why was only this particular item mentioned in the statements? At the very least it demonstrates again the "production line" nature of the statements. At worst, when taken in conjunction with the matters referred to below, it suggests that the tests may not in fact have been "blind" tests.

8.7. Of far greater significance per se, in the Board's view, is the comment in paragraph 5 of each panellist's statement that "When I made my assessment.....I did not know which of the two Sample Lotions was supposed to be superior". That was clearly intended to show the test was indeed a "blind" test but its effect is to suggest the testers knew, in advance of performing their tests, that one lotion was considered by their employer, the patentee, to be superior. Although the choice of assessments they were given was ostensibly threefold (A better than B, A and B the same, B better than A), this advance information effectively ruled out the second option (A and B the same). The evidence thus suggests that the tests were not conducted without an element of bias by advance knowledge.

8.8. With that prior knowledge - that they were expected to find one of the lotions superior - how would the testers behave? The evidence does not say whether or not the testers made their tests together or alone. If they were conducting their tests together (i.e. in the presence of each other), then, after one or two had made their decision, the others would naturally tend to make the same decision. This point increases in importance when one considers the statement of the test organiser, Haruo Ogawa. He says "...the panel members did not know which Lotion was expected to be superior." In other words, the organiser admits, or at the very least infers, the testers knew one of the lotions was expected to be superior.

8.9. In fact, Haruo Ogawa was not just the organiser, he was also one of the testers and made two statements, one in each capacity. His tester's statement is in the same "production line" format as that of the other testers, with the somewhat absurd result that, taking his two statements together, he deposes to having asked himself to conduct the test. Of course, no-one would do that and no-one would intentionally give evidence to that effect but the existence of such an absurdity must further reduce the credibility of the evidence as a whole.

8.10. However, there is a far more important consequence of the two statements by Haruo Ogawa. Since, as organiser of the tests, he knew the compositions of the lotions in advance of the tests (and, therefore, he possibly also knew which was considered, or hoped to be, superior), he could only make one decision as tester. And if, as organiser, his decision was made first and known to the other testers, it is wholly unsurprising, indeed only to be expected, that they unanimously decided A was better than B.

9. In view of the above, the Board has no hesitation in finding that the respondent's latest tests, as presented, are wholly unreliable evidence which must be disregarded.

10. However, the Board also agrees with the respondent (see its letter of 23 May 1997) that the mistakes in the presentation of opponent 2's test evidence (see paragraph 5.3 above), although not so many and so glaring, were such that this evidence should also be treated as unreliable.

10.1. In the case of these tests the table in the test report produced, contrary to the conclusion claimed in the report itself, the somewhat startling result that the patentee's product was considered by the testers to be better. A corrected report was produced in which the original table remained but to which was added a further almost identical table in which those errors had been corrected. The additional table showed the number of persons testing "stickiness" was reduced from twelve to ten and the headings of the columns were transposed so that the Comparative Example 5 lotion and not the Example 1 lotion was shown to be preferred by the testers. No effort was made to repeat the tests, even though the patentee attacked the "doubt" created about these experiments.

10.2. Even if the doubt created about these tests was merely the result of typographical errors, simply amending the report by adding a corrected version of the table was not enough to remove that doubt since, in order to accept the revised evidence as credible, the objective reader has no choice but to accept the proffered explanation (in this case, typographical errors). To do that would entail dismissing any doubt as to the reliability of the evidence which would be unfair to the respondent. Rather, to avoid any objection such as the respondent made, opponent 2 should have repeated its tests and presented the repeated tests in a manner which was not open to question.

11. Both the test evidence of opponent 2 and the respondent's latest test evidence suffer from the same basic defect namely that, for all the tests may in fact have been performed in conditions which ensured the necessary degree of objectivity, the presentation of the results places that objectivity in doubt. This means that the only tests on which the Board can place any evidential value are the tests in the patent itself and the tests reported in the enclosure to the proprietor's letter of 5 July 1996 (i.e. those described in paragraphs 5.1 and 5.2 above).

Article 83 EPC.

12. Claim 1 requires the use of a bivalent salt of an organic acid in a cosmetic composition.

12.1. According to the description of the patent in suit, the bivalent metal salt of the organic acid can be either prepared first and added as such to the cosmetic composition (see example 4), or formed in situ by adding separately the bivalent metal salt and the organic metal salt (see examples 1 to 3).

12.2. These two alternative ways have been described and illustrated in the patent in suit in respectively example 4 and examples 1 to 3. Neither the appellant nor the party as of right have contested the feasibility of this disclosure and the Board has no reason to differ.

12.3. It is indeed true, as pointed out by the appellant in its grounds of appeal, that the skilled person knows that there are bivalent metals which do not dissociate in water, or that there some which are even insoluble. It is also correct that he knows that a weak organic acid might not be dissociated in water (depending on the pH of the solution). The Board moreover agrees that the skilled person is aware that the behaviour of the organic acid and the bivalent metal salt might be influenced by other components present in the cosmetic composition.

12.4. These considerations do however not put into question the fact that the skilled person is able to realise the invention as described in the original disclosure of the contested patent and in particular in the examples. In fact, the only conclusion to be drawn from the appellant's remarks is that there are embodiments which the skilled person would not take into account because he knows that they will not work. There is indeed no point in taking an insoluble bivalent metal salt or an organic acid under such conditions that it would not be dissociated, when the purpose is precisely to form a bivalent metal salt of an organic acid in situ.

12.5. Moreover, the Board observes that such compositions, in which no bivalent metal salt of an organic acid is present, quite simply do not form part to the claimed subject-matter so that the arguments raised by the appellant in this respect appear to be irrelevant.

12.6. Accordingly, as nothing has been added to what has already been considered by the Opposition Division, the Board concludes that its decision that Article 83 EPC was fulfilled is correct.

Article 54 EPC.

13. The appellant has not contested the positive conclusions as to novelty of the claimed subject-matter over the available prior art documents. It has however introduced a new document (5) and alleged that the claim 1 of the patent in suit lacked novelty over the composition described therein containing the following ingredients in water:

.3% ZnO

.0,5% stearic acid

.12% glucose (page 5, example 4).

13.1. The Board however does not agree with the appellant's submission. In fact, ZnO is well-known in the art of cosmetics for its use in the form of a powder and its solubility in water is also well-known to be almost nil as confirmed by the respondent during the oral proceedings. Accordingly, the composition opposed by the appellant does not contain a bivalent metal salt of an organic acid so that this document cannot anticipate the use of a bivalent metal salt of an organic acid in a cosmetic composition.

13.2. Accordingly, the subject-matter of claim 1 of the contested patent fulfils the requirement of Article 54 EPC. The same applies to the subject-matter of its dependent claims 2 to 7.

Article 56 EPC.

14. For the reasons given above (see 2 to 11), the only reliable test evidence is that in the patent and that reported in the enclosure to the proprietor's letter of 5 July 1996. Accordingly document (6), which was filed by the appellant only in relation to the tests filed by Opponent 2 during the opposition procedure, has no remaining relevance.

14.1. This document is moreover silent about the use of bivalent metal salts of an organic acid to suppress the feeling of stickiness in use due to the presence of lecithin in cosmetic compositions.

14.2. Accordingly, as the appellant had no other submissions in its grounds of appeal except its request for an independent expert (see VII), and since the respondent's admissible test evidence demonstrated the effect of the use of bivalent metal salts of an organic acid, the Board can only agree with the Opposition Division's conclusions as to inventive step.