These are partial appeals and cross-appeals from the two judgments (respectively dealing with substantive and further procedural matters) of Pumfrey J of 20th May 2002  RPC 12 and 19th December 2002  EWHC 2772.
The appellants are Reed Business Information Ltd., Reed Elsevier (UK) Ltd. and totaljobs.com Ltd. They are companies within the large multinational publishing group, which goes by the general appellation "Reed Elsevier." For practical purposes in this appeal it is not necessary to distinguish between them. I shall call them collectively "RBI".
For many years RBI have published a big range (about 50) of magazines and journals, such as New Scientist, Farmers Weekly, Motor Trader and so on, some more specialist than others. These carry, in addition to ordinary advertisements for products and services, large sections devoted to job advertisements. So, for many years, job-seekers (as I shall call those wanting employment) have scanned the relevant-to-them sections of RBI’s relevant magazines. Such advertisements are not necessarily placed directly by employers, though many are – they can be placed by employment agencies acting for the employer or more generally by recruitment consultants. Generally speaking if the advertisement is placed by an employer (possibly on advice) he names himself and gives appropriate contact details. On the other hand where there is a middleman who has a financial interest in the placement, the employer is not specifically identified – the job-seeker must contact the middleman and will not know who the employer actually is unless the middleman passes on his name. That he will only do if he considers the job-seeker suitable for the job.
In parallel to RBI’s activities were those of the respondents and their associated companies. Their beginning was in 1960 when Mr Alec Reed rented his first shop in Hounslow. He advertised for secretarial and clerical workers, forming a database of these. Initially he telephoned round employers who wanted staff and put job-seekers he thought were suitable in contact with them. The business expanded geographically and in the kinds of job-seeker and employer for whom it catered. When RBI commenced the activities complained of (September 1999) the respondents had long been a nationwide business covering a wide spectrum of types of jobs and having a physical presence in many a high street. Broadly speaking the business of the respondents can be called that of an "employment agent." I shall call them, as did the judge, "Reed Employment".
Reed Employment registered the trade mark in suit in 1986, shortly after registration of marks for services became possible in this country. The registration is of the word "Reed" as such under No.1,295,450 in Class 35. The specification of services is "Employment agency services included in Class 35."
Class 35 is "Advertising; business management; business administration; office functions." It is currently specified in Schedule 3 to the Trade Marks Rules 2000, reproducing the internationally agreed classification provided for by the Nice Agreement. This vast, very woolly-at-the-edges, class clearly covers the activities of both sides, whether one views the position as at the date of registration (1986), now, or at any time in between. So no point turns on "included in Class 35." But there is a dispute over whether RBI’s activities fall within the words "Employment agency services".
In 1995 Reed Employment first started using the internet as a business tool. By early 1996 it was advertising job vacancies on its website, www.reed.co.uk. The advertisements at that time did not, just as in the case of their paper advertisements (in shops or publications), name the employer. The job-seeker had to go through Reed Employment whose decision it was whether or not to put the two parties in contact.
Version 1 of the totaljobs website
RBI also started to use the internet. It began with on-line versions of its magazines which included job advertisements. It was decided to go further and create a special dedicated jobs-related website, totaljobs.com. The first version was experimental and ran for just a few months, July to October 1999. There followed a series of increasingly sophisticated versions. The Judge describes the various versions in detail. There is no challenge to the accuracy of his descriptions. So I do not need to repeat all the detail. This is particularly so because Mr Geoffrey Hobbs QC for Reed Employment, being a member, as he says, of the "realist school of jurisprudence," has accepted in relation to the issue of whether totaljobs was an "employment agency service" that his case gets no better beyond version 3. It was either such a service by then, or not at all. I think Mr Hobbs was right to make this concession and I am grateful to him for it.
The judge describes Version 1 as follows:
This website was not actively promoted, but it contained links to existing jobs websites associated with RBI publications and directly used the jobs databases of these websites. The only reproduction showing the appearance of this website with which I have been provided is a poor print of the home page, but a number of points can be made:
At the bottom of the page there is a banner advertisement.
The two logos look like this:
I would add two things. First, that the banner advertisement adds nothing relevant and secondly that there was the following message adjacent to the RBI logo: "Contact Reed Business Information if you would like to advertise your company’s job vacancies."
The judge summarised the position thus:
It seems to me clear that Version 1 would appear to the user as a Reed Business Information website. Although the title of the site is clear enough, the references to Reed Business Information as both the contact point, as the supplier of magazines to which a subscription may be purchased and as the subject of the ‘about ....’ make this conclusion inevitable.
Mr Martin Howe QC for RBI does not challenge the judge’s summary of the position – Version 1 would be perceived as emanating from a business called "Reed Business Information".
The General Nature of the Issues
All versions of the totaljobs website which contained a use of the word "Reed" (visible or in metatag form) were accused of constituting passing off and infringement of registered trade mark. The Judge found both. Largely he dealt with all versions together. He dealt with infringement first. It is not entirely clear whether he held there was infringement pursuant to s.10(1) of the Trade Marks Act 1994 (implementing Art. 5(1)(a) of the Trade Marks Directive (89/104)). He clearly held there was at least infringement pursuant to s.10(2) (implementing Art.5.1(b)) and, save in relation to very minor matters (use on invoices and at a trade show) that section 11(2) (implementing Art. 6.1) afforded no defence. As to passing off he again dealt with all versions globally, holding that "the classical trinity of ingredients in a passing off action, reputation, confusion leading to deception and damage, are present" and that save in respect of the same minor matters no defence of own name applied.
I think this "overall" approach in relation to trade mark infringement can lead to muddle and error. It is essential to know in respect of each alleged infringement whether it is an Art. 5.1(a) or 5.1(b) case. If it is the former one need go no further save to consider whether an Art.6 defence applies. If it is the latter then there is a vital "likelihood of confusion" question to be answered. For that reason I intend to be rather more systematic than the Judge.
RBI do not challenge all of the Judge’s findings. They dropped the use of both logos in June 2000 (the Reed Elsevier logo after Version 1 and the RBI logo during the course of Version 4) and do not contest a declaration in the following terms:
IT IS DECLARED
The passages in italics are those which RBI wish to have inserted into Pumfrey J’s order on this appeal. They seek a similar modification to the order for the inquiry as to damages.
What gives RBI particular commercial concern is the effect that the judge’s unamended declaration may have on their freedom to use the copyright notice, and the scope of the inquiry as to damages. In relation to the latter Reed Employment are seeking a substantial sum even though (thus far at least) they can point to no diversion or diminution of their trade caused by any of the activities held to have been unlawful – they claim damages on a user basis as more fully discussed in Pumfrey J’s second judgment.
Subject to the Art 6.1 defence, in order for the copyright notice to have a chance of escaping infringement it is necessary for its use to fall outside Art. 5.1(a) If it falls within then there is a straightforward and unanswerable same services/same sign case which could only be saved by Art.6.1. Thus it is that much of the debate before us, even in relation to the now-obsolete, short-term and hardly used Version 1, focused on whether there was Art. 5.1(a) infringement. It gave rise to the slight oddity that, although infringement in respect of the logos was not contested, its legal basis was. Since there could be important commercial consequences for RBI (and its parent, Reed Elsevier) depending on the answer, it is legitimate for this question to be examined.
Before proceeding to do so, I set out the two infringement provisions in play here. I take them from the Directive because it makes this judgment more intelligible to a reader in another EU country, our Parliamentary draftsman having unhelpfully implemented the Directive verbatim but with re-numbering:
Identical mark and sign?
To bring a case within Art.5.1(a) the proprietor must show two things, (i) use of a sign which is identical to his registered mark, and (ii) use for goods or services which are identical to those for which his mark is registered.
So is this a case of "identical sign"? The defendants have never used "Reed" alone (save for the minor Yahoo use dealt with below). They used "Reed" in the logos and either as part of the composite "Reed Elsevier" or (and much more frequently) as part of the composite "Reed Business Information", sometimes with and sometimes without the addition of "Ltd."
Since Pumfrey J’s judgment the ECJ has spoken in relation to "identical marks." It did so in LTJ Diffusion v Sadas Vertbaudet Case C-291/00 ("Arthur" and "Arthur et Félicie"). The plaintiffs’ mark was a device consisting of the word "Arthur" written in a fancy script in the nature of a signature. It looked like this:
The defendants too used a device mark:
[The pictures are taken from  FSR pp.612-3.]
The referring court asked "whether the prohibition in Art 5(1)(a) covered only identical reproduction, without addition or omission, of the sign or signs constituting a mark, or whether it could extend to (1) reproduction of the distinctive element of a mark composed of number of signs; (2) full reproduction of the signs making up the mark where new signs were added."
The ECJ said:
This guidance is opaque. On the one hand the criterion of identity "must be interpreted strictly." On the other the court is to allow for the "imperfect picture" of that notional construct "the average consumer." This latter is a concept more aptly related to the question of whether there is a likelihood of confusion between an accused sign and the registered mark, namely the Art. 5.1(b) question.
Of importance here is the recognition that an addition in the defendant’s sign to a registered mark may take the case outside one of "identity" (see para. 51). This is obviously sensible – one word can qualify another so as to change its impact, "Harry" qualifies "Potter" and vice versa, for instance. It is particularly in the recognition that additions can change identity that the ECJ has moved on from the rather rigid view taken under the old UK law.
As a matter of policy there is no reason to suppose that the Court meant to soften the edges of "strict identity" very far. For even if a sign and mark are not "identical" for the purposes of Art.5.1(a) if there is a likelihood of confusion it will be caught by Art.5.1(b). The reason for the differences between the two provisions is at least partially set out in the 10th recital
Whereas the protection afforded by the registered trade mark, the function of which is in particular to guarantee the trade mark as an indication of origin, is absolute in the case of identity between the mark and the sign and goods or services; whereas the protection applies also in case of similarity between the mark and the sign and the goods or services; whereas it is indispensable to give an interpretation of the concept of similarity in relation to the likelihood of confusion; whereas the likelihood of confusion, the appreciation of which depends on numerous elements and, in particular, on the recognition of the trade mark on the market, of the association which can be made with the used or registered sign, of the degree of similarity between the trade mark and the sign and between the goods or services identified, constitutes the specific condition for such protection;
Advocate-General Jacobs’ opinion makes it clear that, save for the plaintiff, no party (the defendants, the Commission, the French and UK governments) contended for anything other than a strict interpretation (see paragraph 32)). Moreover, although there is nothing express in the 10th recital about a danger of confusion being presumed in the case of identity, both the travaux préparatoires of the Directive and Art 16(1) of TRIPS are explicit on the point (see Jacobs A-G at paras. 34 and 35).
The Advocate-General’s opinion is also helpful in indicating the sort of difference which would be so minute as to leave the mark and sign "identical" and that which would not. He said, para. 50:
I suggest that the reproduction of [the plaintiffs’] mark in the same distinctive script but without the dot under the initial ‘A’ might well have been perceived by the average consumer as identical to the original (the change being minute and wholly insignificant) whereas the use of a noticeably different script and/or the addition of another name might be seen as only similar (such changes, at least taken together, being substantial).
This approach also makes sense if one considers the registered trade mark system as a whole. To be registerable a mark must have a "distinctive character" (see Art.3(1)(b)), a matter which depends partly on the nature of the mark itself and partly on whether it has in fact acquired such a character (see Art.3(3)). So the registration authorities (either national or in OHIM) must have regard to the nature of the mark applied for. In the case of a device mark consisting of a word written out in some special way that will include its visual impact. It would make no sense if, once such a mark were registered, a visually different but identical word device mark were treated as identical.
Nor do I think it makes any difference to the identity question that trade mark use includes spoken use. Section 103(2) of the Act explicitly makes that clear, providing that reference to "use" "includes use .... otherwise than by means of a graphic representation." Although there is no corresponding provision in the Directive it seems tolerably clear that it is implicit that purely oral use may count for infringement. This is because Art.5.1(3) provides a non-limiting definition of use which may infringe which covers pretty well any form of visual use one can imagine. It says:
It is implicit from this that the concept of use is very wide and must surely cover oral use.
It follows, I think, that in the case of any accused sign which has some visual significance, one must take that into account in considering the question of identity. The visual impact may be slight, but it will count even if the marks are identical to the ear. You can only have "identity" if there is both aural and visual identity.
But before I get to the question of identity here, there is a further preliminary matter. What is to be compared with the registered mark for the purposes of Art.5.1(a)? It is the defendant’s "sign". So this must be identified first. In many cases this is easy – the defendants’ sign in "Arthur et Félicie" is an example. And during the registration process, where there is a potential conflict with an earlier registered mark this preliminary question is self-answering. The fact that what the defendant’s sign is is often so obvious that it does not merit a thought, does not mean that one can overlook the preliminary question. There may be other cases where some difficulty is posed, particularly where descriptive or semi-descriptive words are added to what is obviously a word mark.
Although Mr Hobbs accepted that the defendant’s sign must first be identified, he overlooked that when he relied upon the BMW case (BMW v Deenik Case C-62/97  ECR-I 905). In that case the expressions complained of were "BMW specialist," "Specialised in BMWs" and "Repairs and maintenance of BMWs". The Court proceeded on the basis that this was an Art.5(1)(a) case of identical marks and goods. It did not explicitly first identify the sign used by the defendant. But it was obvious what it was: just BMW. The other words in context were wholly devoid of any trade mark significance. So I do not think BMW is relevant on the question of identity of mark and sign.
In the present case it is accepted by Mr Hobbs (rightly I think) that the defendants’ signs are the two logos, the composite words "Reed Elsevier" and the composite words "Reed Business Information" with or without the word "Ltd.". It is each of those signs which must be compared with the mark consisting of the word "Reed" when considering the question of identity.
So far as the logos are concerned Mr Hobbs did not contend for identity. Nor do they matter, given that there is no appeal in respect of their use. Mr Hobbs also realistically accepted that there was no identity in respect of the words "Reed Elsevier". This phrase is a good example of an addition that makes a difference.
It was over "Reed Business Information" that battle was joined. The composite is not the same as, for instance, use of the word "Reed" in the sentence: "Get business information from Reed." In the latter case the only "trade-marky" bit would be "Reed". In the former, the name as a whole is "Reed Business Information." The use of capital letters is of some visual significance – it conveys to the average user that "Business Information" is part of the name. If the added words had been wholly and specifically descriptive – really adding nothing at all (e.g. "Palmolive Soap" compared with "Palmolive") the position might have been different. But "Business Information" is not so descriptive – it is too general for that.
So is "Reed Business Information" identical to "Reed"? I think not. "Reed" is a common surname. The average consumer would recognise the additional words as serving to differentiate the defendant from Reeds in general – this one calls itself "Reed Business Information" because it supplies information to businesses in some unspecified way or ways.
Putting it another way, I do not think the additional words "Business Information" would "go unnoticed by the average consumer." In all uses of the phrase complained of they are as prominent as the word "Reed."
In so holding I am not saying that in some circumstances the average consumer could not assume that "Reed Business Information" is connected with Reed Employment or an organisation called "Reed." But these would be cases of similarity of mark and sign, not identity.
It follows that in all cases complained of there is no identity of mark and sign.
Identity of Services?
The mark is registered for "employment agency services." RBI say that the totaljobs website has never offered such services. This depends upon what the specification means. An argument raged over two points. The first was whether the specification should be construed as of 1986 (the date of registration or filing, which by s.40(3) are the same thing) or the date when the defendant commenced its activities. The second point was whether the services fell within the specification on either date.
Before turning to these points it is as well to consider the general principles as to the construction of a specification of services. In Avnet v Isoact  FSR 16, I said that:
specifications for services should be scrutinised carefully and they should not be given a wide construction covering a vast range of activities. They should be confined to the substance, as it were, the core of the possible meanings attributable to the rather general phrase.
Neither side dissented from this. The proposition follows from the inherent difficulty in specifying services with precision and from the fact that a service provider of one sort is apt to provide a range of particular services some of which will be common to those provided by a service provider of another sort. Here, for instance, both sides publish advertisements for jobs and have done so for years. No-one who has looked into a Reed Employment high street shop could have missed these. Nor could anyone have missed RBI’s job advertisements in their various magazines.
Accordingly I think that principle applies here. What one must do here is to identify the core activities which make a service provider an "employment agency."
Before coming to that I must deal with the date point. I do so briefly because I do not consider that the case turns on this. Mr Howe contends that a specification of goods or services cannot change its meaning with time. Mr Hobbs submits that it can – if the "core" nature of the service changes with time, then the meaning changes too.
I have no doubt that Mr Howe is right. One can test the point best by reference to a registration qualified by the words "included in this Class". From time to time, though rarely, the class in which a particular kind of article is put is changed by international agreement. If that happens it is inconceivable that the trade mark owners’ rights could be changed. "Included in Class X" must mean "included at the time of registration."
So also for a word or phrase which changes its meaning over time. But that must in practice be very rare. Indeed I know no instance of it in any reported case. The ordinary case – and I think this is one – is where some new variant of an article or service comes into existence after registration. The issue then is whether that new article or service falls within the meaning of the existing specification. Columbia Graphophone’s TMs (1932) 49 RPC 621 is a good example of a new article falling within an old specification. The specification of goods of the mark under attack was "all goods in Class 8". Class 8 (of the old classification, bizarrely taken from the classification used for the Great Exhibition) quaintly read "philosophical instruments, scientific instruments and apparatus for useful purposes; instruments and apparatus for teaching". An application for partial rectification was made on the grounds of non-use. The excision sought was essentially for "cinematograph films, talking and silent" (my précis). The trade mark owner, a record company, sought to justify the non-user by saying that talkie films had only just become possible and that amounted to special circumstances. The argument failed and excision was ordered. There was an express excision from the specification of a kind of article (talkies) that did not exist at the time of registration.
Mr Howe’s position is further supported by the authorities, although to be fair the issue of a change of meaning of a specification over time was not directly in issue in any of them. It was just assumed – because it was so self-evident - that the meaning was the meaning at the time of registration, both in pre- and post-1994 Act cases. For that reason I merely give the references and do not overburden this already long judgment with citation. The cases are GE TM  RPC 418 at p.458; and my own decision in British Sugar v James Robertson  RPC 281 at pp.289-90. The learned editors Kerly 13th edn., 2001 are of the same opinion (paras. 4-04, 13-42).
The judge refers to neither of these cases or Kerly. He said that:
It seems to me that the nature of the register requires the relevant date to be the date when the acts complained of started. If at that date the defendant had looked at the register and asked itself ‘is what I am about to do the provision of an employment agency service’ the register would have done its job.
With respect I do not follow this. Why not pick a later date? Suppose the meaning expanded after the defendant started his activities? Could a defendant who started outside the specification find himself moved into an Art.5.1(a) position by changes in the meaning of language? And would defendants who started at different dates potentially stand in different positions? I think the answer to these questions is obviously not.
However I do not think the point arises on the facts of this case, to which I now turn. Reed Employment’s expert, Miss Little, identified 13 activities "all or some of which" would be offered by an employment agency. Her list does not include any activity that can only be done through the internet. This to my mind makes it clear that there has in fact been no change in meaning over the years since registration. Indeed she said that in her opinion the components of her list have not altered materially – only the manner of delivery has changed. Further, Mr Hobbs invited us to consider whether the activities of RBI, if carried out manually, would amount to that of an employment agency. That implicitly recognises that the question of employment agency or not is not one which turns on the internet based nature of the services offered or upon any change in meaning over the years since registration.
So I think the question is not whether "employment agency services" has changed its meaning over time because of the advent of the provision of services over internet. The question is whether the services provided over the internet by RBI are those of an employment agent.
This requires an inquiry as to what the core features of an employment agency are. What is it that distinguishes an employment agent from others who provide services in the recruitment industry? Both sides called experts who gave evidence as to what organisations who call themselves "employment agents" do. There was also some indirect evidence derived from members of the public who took part in surveys conducted for RBI (the "Serco" surveys) of Versions 1 and 4 of the totaljobs website for the purpose of evaluating them. The responses give some, albeit incidental to the purpose of the survey, indication of the public’s idea of what they expect from an employment agency.
Mr Howard, RBI’s expert, commented on the 13 activities identified by Miss Little, pointing out that a number of the activities were not exclusive to an employment agency service. Thus for instance, one item mentioned by her is "Advertising of vacancies in media which are likely to attract suitable candidates." This can be and is done by employment agencies, but it is not an activity exclusive to such agencies – the employer can advertise directly himself, for instance. So also for other identified activities such as "dealing with advertisement response", "assisting candidates with the production and content of their CVs and submitting these to potential employers," "assistance with reference checking of candidates" and others. These are not the touchstone of an "employment agent".
In the end, therefore, Miss Little’s list is of little assistance. It is not without significance, however that none of the activities identified was available as such on any version of the totaljobs website.
That is of particular importance in relation to the three particular activities within the list which Mr Howard accepted are unique to employment agents – "contacting potential employers about candidates’ availability", "advising the client (i.e. the employer) and candidate (job-seeker) at an employment offer stage" and "for temporary staff, carrying out payroll and other HR (human resources) management activities." Of course the absence of such an activity by totaljobs is not conclusive, but it is a strong pointer that something other than a service of an employment agent is being provided.
In the end therefore, Reed Employment’s case had to be and was more general than one depending on particular activities. Miss Little put it this way:
.... employment agency services’ cover all services for candidates or clients which play a part in the recruitment process.
But that must be too wide – it would cover RBI’s traditional business of publishing job advertisements in magazines for instance.
Mr Hobbs put the test more narrowly: he submitted that the provision of "employment agency services" involves the finding of suitable candidates for jobs and/or the finding of suitable jobs for candidates. He likened the service to that of a marriage broker – the essence of an employment agency was matching job-seekers with employers.
The Judge accepted this submission. He said:
I have found it surprisingly difficult to come to a clear conclusion. I think that the better view is that the services provided by the defendants are employment agency services because totaljobs.com acts as a go-between between clients and candidates. I accept that the services do not include the matching by an independent consultant which one would expect from an employment agency (or at least the claimants) and that the element of judgment is not present. But totaljobs.com does enable the client to establish a set of criteria to apply to those who respond to its advertisements. I am conscious also that employment agencies do use totaljobs.com to obtain candidates, and this suggests that its services as an advertiser are more important than its other services. I think that this is not sufficient to outweigh the other factors which I have identified.
Mr Howe attacked this. First he pointed out that immediately prior to that the Judge adopted a narrower meaning. The Judge said:
Finally I should refer again to the anonymous user who was quoted in the usability evaluation of totaljobs.com which took place in September 1999.
This quotation encapsulates the evidence of the expectation among candidates of what an employment agency provides and thus the core of what is meant by ‘Employment agency services’ in this context.
I think Mr Howe is right – the core meaning adopted here involves a vetting for suitability by the agency – which does not happen with totaljobs.
Mr Howe further submitted that the Judge entirely overlooked the word "agency". RBS were providing a publishing service – updated in its method of delivery and the fact that it was searchable in a much more efficient manner than was possible with mere paper publication, but essentially only a publishing service.
I agree. I start with Version 1. This carried job advertisements placed by clients. What went into the advertisement was a matter for the client – he could identify himself and would naturally do so if he were the employer. If the advertisement were placed by an employment agency however, the employer would not be identified – for it is essential to the remuneration of such an agency that it and it alone puts the job-seeker and employer in contact. It gets paid on the basis of job placements. The Version 1 site was searchable by the job-seeker in that he could specify criteria for jobs he was looking for (kind of job, area, pay and so on) and the computer would find all the job advertisements complying with his criteria. But I do not see that the provision of the search facility made RBI an agent – even in a loose sense – for the job-seeker. Nor did RBI act as an agent, in a loose sense, for the employer save in that it put the employer’s advertisement on its site – in other words published it there. Whether the employee got in contact with the employer would not affect RBI’s remuneration – indeed it would not even know whether any jobs were landed by use of the site. Following Mr Hobbs’ analogy it would be a very odd marriage broker who did not and could not know whether and if so how many marriages came about as a result of his activities.
The public reaction to Version 1 confirms this. The Serco study involved taking a dozen readers of various RBI magazines (i.e. potential job-seekers), letting them experiment with the site (including carrying out of set tasks) watching them through a one-way mirror and finally asking for their reaction. A number of them thought the site was from an employment agency. That in turn led to an expectation of what the site would do. Thus the Serco study says:
Even participants who did not link totaljobs to Reed Employment thought that the site was provided by an agency. One participant, a recruitment consultant, was very surprised to see direct contact information being displayed: ‘I have never seen a site giving company information away.’
That is important – the participant was noticing that direct contact without the involvement of an agent, was possible on this site. A true employment agency would never let job-seekers make direct contact with its employer clients. Its list of clients is a crucial trade secret. It would be out of business if it published their names because job-seekers and clients would simply make direct contact.
Another unmet expectation was in relation to what the site would do for job-seekers – I have already quoted the relevant passage in the quotation from the judgment but requote it here.
Some participants expected the site would provide them with access to recruitment consultants, and this was most evident in their expectation of being able to specify their skills and requirements and that CVs could be submitted to an agent who ‘won’t even put you forward if you’re not suitable.’
This is clearly saying the site does not do what an agency does. This is highly significant. The core of an employment agency service lies not merely in the fact that job-seeker and employer can be matched. Matches can be made in a variety of ways: for instance the job-seeker may read an advertisement placed by an employer and respond to it directly. No-one would call the publisher of the advertisement an agent in those circumstances, even though he has played a part in the matching. Playing any sort of part in a matching process is not enough.
The key core elements of the service of an employment agent are, it seems to me that the agent:
Has control over the introduction between employer and job-seeker; and
Has a direct interest in the match being made – as Mr Reed put it "we do not just supply warm bodies."
It is this interest, which gives the employment agent a real incentive to supply only good people to employers – it is on the basis of his performance of this key task upon which the reputation of an employment agent rests. This last point is of course of particular significance in a trade mark context – which is the relevant context for considering a specification of goods or services.
I conclude that Version 1 was not an employment agency service. It was in substance a searchable jobs advertisement site and no more.
I turn to Version 3, the latest version I need consider in view of Mr Hobbs’ acceptance that if totaljobs was not an employment agency by then, it never was. By this time a variety of functions had been added. Those suggested to be of relevance were a CV database and a jobs-by-email service.
For the first time job-seekers could put their own information onto the site. They could put their CVs onto a database within the site. Job-seekers could also, if they so wished have their CVs sent to interested recruiters.
The jobs by email service ran like this. The job-seeker could specify criteria about the job he was interested in. The site was programmed then to send him by email details of up to 25 jobs complying with the criteria, the search for these being run over the new jobs which had put onto the site following the request.
Did the addition of either of these facilities turn the site into that of an employment agency? I think not. There was still no element of control by totaljobs of what was going on and no direct interest in matching or responsibility for a match. Yes, the facility made it easier for employer to find job-seeker and vice versa, but the match was entirely up to them. All the site did was to provide information – with no responsibility for it in law or as a matter of commerce. An employer would not blame totaljobs if an employee found through it proved unsuitable. He would not say, "totaljobs has let me down." He might say "totaljobs is not a good way of finding employees" but that is a different thing altogether – he would not regard totaljobs as his agent.
Mr Hobbs suggested one way of testing the position was to imagine what totaljobs did being done manually. That test does not support him. Consider a job-seeker looking for a job having certain criteria set by him. The manual equivalent would be letting him search manually through all the advertisements "inserted" by employers into totaljobs files to find those which fitted his criteria – just like scanning a paper publication. The manual equivalent of the CV forwarding service would be for him to send his CV to those employers he found in the files who satisfied his criteria. The 25 jobs daily email is the manual equivalent of a daily search for the desired criteria through the new job advertisements. An employment agency would allow none of these things.
I should mention two other matters. Mr Hobbs relied upon the statutory definition of an "employment agent" in s.13 of the Employment Agencies Act 1973. This reads:
For the purposes of this Act ‘employment agency’ means the business .... of providing services (whether by the provision of information or otherwise) for the purpose of finding workers employment with employers or of supplying employers with workers for employment by them.
But the purpose of this legislation is far removed from identifying the core nature of the service for the purposes of a trade mark monopoly. The Act is for protecting job-seekers from unscrupulous people who offer to get them jobs. It is simply irrelevant to the current problem.
The second point is that in 1999 RBI applied to register "totaljobs" as a Community Trade Mark with a specification for services which included "employment agency services". The inference Mr Hobbs suggested we should draw is that RBI themselves considered that "totaljobs.com" was offering such a service. This is desperate stuff. Anyone with knowledge of the practice of OHIM (the Office for the Harmonisation of the Internal Market) will know that it does not require any statement of an intention to use the mark applied for. It permits specifications of goods and services of any width chosen by an applicant – who has no motive for restricting that width and every motive for keeping it wide. Moreover RBI accept that one activity, carried on experimentally under the name totaljobs (without any use of the name "Reed") for a brief period from May 2001, was that of an employment agency service. So the fact they applied for registration with such a specification cannot mean that implicitly they accept that their activities were such services.
Art. 5.1(b) – the principles
So there is neither identity of mark and sign nor identity of services. This is not an Art.5(1)(a) case. All the uses in issue on this appeal must be judged by Art. 5.1(b). I turn to the applicable principles under this provision.
The ECJ has considered these in a series of cases. The Judge helpfully said:
the nature of the comparison, and the factors to be taken into account, have been considered by the ECJ in a number of cases. The cases on the corresponding provisions of the Directive, that is, Art 5(1) and 5(2) are: Case C–251/95 Sabel v Puma  ECR I-6191, Case C–39/1997 Canon v MGM  ECR I-5507,  RPC 117 Case C–342/97 Lloyd Schuhfabrik Meyer & Co GmbH v. Klijsen Handel BV  ECR I-3819,  ETMR 690 and Case C–425/98 Marca Mode CV v Adidas AG  2 CMLR 1061. From these cases I derive the following propositions:
Neither side criticised these propositions. I have a few further observations. First, as to proposition one. The Court has laid down the "global assessment" test (Sabel para. 22, Lloyd para. 18). This requires the court to take all the circumstances into account. As the Judge observes this involves many of the factors involved in passing off. But there is this difference: in the case of passing off the court has regard only to the actual reputation and goodwill of the claimant at the time the defendant starts his activities. In the case of Art. 5.1(b) infringement the court must assume, even if it be not so, that the mark owner has a reputation and goodwill in his mark. In other words the court must assume that the mark has been in use and developed such a reputation and goodwill for the specification of goods or services.
Were this not so, there could never be Art. 5.1(b) infringement of an unused registered mark. That would defeat one of the key purposes of trade mark registration – the conferment of protection in advance of use, before a reputation under the mark has been developed. For once there is a reputation most (probably all – I have never heard of one that does not) systems of law confer protection independently of registered trade marks by some sort of unfair competition rules. In the UK that is by the law of passing off.
The way one arrives at this facet of Art. 5.1(b) infringement is via the word "likely". This implicitly directs attention to the future – which includes the assumption that the mark owner will put his mark into use if he has not done so yet. The kind of use contemplated when a mark is unused or minimally used must be that of "notional and fair" use – no other makes sense. When however the mark sued upon has actually been used on a substantial scale, one can take that actual use into account in assessing its distinctiveness (which depends on a combination of its inherent nature and its factual distinctiveness) and the consequential "global assessment." That applies here.
Next the ordinary consumer test. The ECJ actually uses the phrase "average consumer" (e.g. Lloyd paragraphs 25 and 26). The notion here is conceptually different from the "substantial proportion of the public" test applied in passing off (see e.g. Neutrogena v Golden  RPC 473). The "average consumer" is a notional individual whereas the substantial proportion test involves a statistical assessment, necessarily crude. But in the end I think they come to the same thing. For if a "substantial proportion" of the relevant consumers are likely be confused, so will the notional average consumer and vice versa. Whichever approach one uses, one is essentially doing the same thing – forming an overall ("global") assessment as to whether there is likely to be significant consumer confusion. It is essentially a value judgment to be drawn from all the circumstances. Further conceptional overelaboration is apt to obscure this and is accordingly unhelpful. It may be observed that both approaches guard against too "nanny" a view of protection – to confuse only the careless or stupid is not enough.
Finally, although I agree with the Judge’s questioning of the Court’s proposition of fact that "there is a greater likelihood of confusion with very distinctive marks" there is some truth with the opposite proposition. The Court in Lloyd said:
In determining the distinctive character of a mark, and accordingly in assessing whether it is highly distinctive, it is necessary to make a global assessment of the greater or lesser capacity of the mark to identify the goods or services for which it has been registered as coming from a particular undertaking, and thus to distinguish those goods from those of other undertakings. In making that assessment account should be taken of all relevant factors and in particular of the inherent characteristics of the mark, including the fact that it does or does not contain an element descriptive of the goods or services for which it has been registered.
The last sentence is an acknowledgement of a fact that has long been recognised: where a mark is largely descriptive "small differences may suffice" to avoid confusion (per Lord Simonds in Office Cleaning Services v Westminster Window and General Cleaning (1946) 63 RPC 30 at p.43). This is not a proposition of law but one of fact and is inherent in the nature of the public perception of trade marks.
It is worth examining why that factual proposition is so – it is because where you have something largely descriptive the average consumer will recognise that to be so, expect others to use similar descriptive marks and thus be alert for detail which would differentiate one provider from another. Thus in the cited case "Office Cleaning Association" was sufficiently different from "Office Cleaning Services" to avoid passing off.
The same sort of consideration applies when there is use of two common surnames, as in this case. The average consumer will be alert for differences – just in the same way as one distinguishes WH Smith from other Smiths by the initials. That is of importance here in making the global assessment.
The estoppel point
Before applying the Art. 5.1(b) test to the present case I must consider Mr Hobbs’ estoppel point. It is based on the fact that RBI do not appeal Pumfrey J’s order so far as it relates to the logos. Mr Hobbs contends that it follows that:
there is an unappealed finding that the RBI and Reed Elsevier logos are likely to cause confusion,
the sole reason for that is the use of the words "Reed Business Information" and Reed Elsevier.
it follows there has been a judicial determination that those words are likely to cause confusion,
from that there has been no appeal,
so it is not open to RBI to contend otherwise on this appeal.
This is fallacious for a number of reasons. First, the only determinations about the logos have been in the specific contexts of the versions in which they were used, namely Version 1 (both logos along with a clear statement next to it telling readers to contact "Reed Business Information" if they wished to advertise), and Versions 2–4 (RBI logo only). There is no general determination that any use of the words "Reed Business Information" in any context is confusing. Secondly it is not clear that there was a judicial determination of infringement pursuant to Art.5.1(a) so Mr Hobbs cannot rely on any finding that there was. Thirdly the rule as to res judicata can only bite when there has been no appeal. But this is an appeal –so to the extent that any point is raised here it cannot be the subject of an estoppel – this appeal trumps the estoppel.
I am not sorry to reach this conclusion. The point is one of the utmost technicality. If right it would mean that RBI should have appealed everything, even though they have no interest in using either logo for totaljobs.
Application of Art. 5.1(b) to the various uses
Before embarking on a consideration of these it is important to appreciate that there has at all times been some harmless confusion between the two sides. This is hardly surprising given that they both use the common surname Reed and they both have something to do with jobseekers finding jobs and employers finding jobseekers. The confusion has essentially been of the "I wonder if there is a connection" or "are they sister companies?" kind. The Judge elevated this into some sort of principle at the outset. He said:
I do not think this is an accurate statement of the law. It draws a distinction between confusion and damaging deception that in the real world is elusive. Secondly it imposes a wholly impractical requirement – reduction of a risk of confusion to zero when the parties’ names are common. The test is not that of passing off or trade mark infringement – it is some sort of super-added test which, if it goes further than that the claimant must prove passing off or trade mark infringement, has no warrant.
It follows that what must be done here when considering infringement is simply to apply the Art.5.1(b) test to the various uses under appeal. I should say here that whilst it is legitimate to consider such uses in combination where that has occurred (a "global assessment"), it does not follow that that which is confusingly similar in combination with other matter must be regarded as so in isolation.
Use of the copyright notice "© [year] Reed Business Information Ltd."
The notice was used continuously from the outset on Version 1 to part of the period of Version 5, use ceasing on 3rd August 2001. The only purpose of the use was to make an explicit claim to copyright. Although this is not required to ensure copyright protection under UK law or those of other Berne Convention countries, such a claim is necessary for US (and possibly other) laws. The use was not intended to have the effect of designating origin and was kept as minimal as possible, serving as it did, no trade or promotional purpose. For much of the time the use was at the bottom of the homepage which in practice meant that the user had to scroll down to see it at all. It would have been no interest to either jobseekers or employers, the "average consumers" of the site.
Mr Howe submits that there is simply no evidence supporting the conclusion that this use, considered on its own, was likely to confuse. He is right about this in the sense that there is no direct evidence about the notice alone. All the court has to go on is inference from other material, the inherent nature of the mark Reed, the nature and form of the uses complained of and the nature of the parties’ respective services.
I start with the last. As I have held the services are not the same. But they are "similar" in that they are closely related – the sort of service that a jobseeker or employer might use. Mr Howe did not seek to argue non-similarity. So the degree of similarity of the services is a factor to be taken into account – see Canon.
As to the inherent nature of the mark and sign complained of, it is relevant to take into account that they involve the common surname Reed. As I explained above it follows that small differences may suffice to differentiate.
As to the general factual position, the Judge found that Reed Employment’s mark "Reed" was well-known amongst the general public (and particularly jobseekers). RBI, he found, was also very well-known to employers, advertising agents and the like, specialists in the recruitment business generally, but was "not so well-known" to the public.
And as I have said the Judge also found that there had long been a general background of unimportant confusion arising from the fact that they operated in the general field of recruitment services and had similar names.
The Judge did not have any direct evidence from the public at all. Nor did Reed Employment, who have much direct contact with jobseekers all around the country, produce any evidence that any of its staff had ever come across anyone who thought that totaljobs was a Reed Employment website. This was despite the fact that members of Reed Employment’s online team were told to be on the lookout for it. The nearest to direct evidence of members of the public, and the material upon which the Judge relied as evidence of confusion, consists of the two "Serco" studies.
The first of these was in relation to Version 1. Version 1, as the judge held, presented itself as emanating from "Reed Business Information." Serco concluded that most participants "seemed to understand that RBI were the owners of the site, but thought they were part of Reed Employment." It was on the basis of this internal evidence that the Judge found that there was passing off and at least Art.5(1)(b) infringement. It is, I think, of some importance to note that the individuals who thought that this site was from Reed Employment did so spontaneously. The participants were not prompted.
The Judge’s conclusion in relation to Version 1 is therefore understandable, as is the fact there is no appeal about it. But it is essential to bear in mind that the participants’ reactions were based on the site as a whole – the logos, the copyright claim and most obviously of all the explicit statement inviting those who wanted to advertise to go to Reed Business Information.
By the time of the second Serco survey, on Version 4, use of "Reed" had diminished considerably. Only the logo and copyright notice remained. The logo was on the home page as it came up on the screen, as was the copyright notice, albeit in small print. Serco said this, in relation to branding:
Only one participant commented spontaneously on the Reed Business Information logo and origin of totaljobs. When participants were asked who was responsible for the site, most had ‘no idea’ at first but scrolled quickly to the bottom of the page to find out.
While participants often did not know who had developed together [this must be a error for "the site"], this did not appear to have a negative effect on their trust or perceptions of the site. Few users questioned who was behind the site, and some said they trusted the brand [i.e. "totaljobs"] instinctively because they had seen it advertised.
Having seen the Reed logo, some participants believed that Reed Employment was responsible for totaljobs. This gave one participant the impression that the site would be biased towards secretarial work.
Serco concluded that RBI should:
reduce the emphasis of the Reed Business Information logo, or perhaps remove it altogether.
Maintaining the copyright information in its current position would provide some indication of the origin of the site to users who are concerned about it.
So such evidence of confusion as there was emerged only after questions of the sort deprecated in opinion polls conducted for the purpose of providing evidence of confusion in trade mark and passing off cases (see the well-known observations of Whitford J in Imperial Group v Philip Morris  RPC 293). I particularly have in mind his observation at p.303:
Great importance inevitably attaches to the way in which the questions are cast. It is very difficult in this connection with an exercise such as this to think of questions which, even if they are free from the objection of being leading, are not in fact going to direct the person answering the question into a field of speculation upon which that person would never have embarked had the question not been put.
The second Serco study involved just such prompted speculation – to the question who was behind the site. And even then the level of "confusion" was low. It is fair to state that RBI reacted to the suggestion of confusion by taking it seriously (see the internal emails recited in paras. 67,68 and 70 of the 1st judgment. But the Judge did not and could not have made any finding based on these emails, which are not themselves evidence of confusion.
There was no direct evidence indicating any confusion arising from the use of the copyright notice alone (i.e. without any other use of "Reed" on the website"). It was used alone for about a year, so the issue is live. But the Judge did not make a direct finding of confusion arising from the use of the notice alone. In the circumstances we have to make up our own minds on the point. Can confusion or its likelihood be inferred from other evidence? I think not. It is for the claimant to prove its case. I think the material is too thin to justify the inference.
I say this because the nearest (where there was any evidence of confusion) to use of the notice alone was Version 4. For that the RBI logo was used along with the notice. The totality of this evidence was the second Serco study, of which I have set out all the relevant parts. That evidence is very thin. It remains a pure matter of speculation as to whether the notice alone would even have this thin effect – which was more probably the result of the use of the logo than the notice. For the logo might just (in context) give the idea that "Reed Business Information" was behind the site – though even this involved prompted speculation. A copyright claim is not an explicit indication of who owns the site. What the average consumer understood by it (assuming the average consumer would have paid attention to it at all) remains completely uncertain. All that is known is that a further, unprobed, study done during the period of use of the notice alone did not throw up any evidence of confusion. Infringement in respect of the notice alone is not made out.
Other use of Reed
By their written submissions following provision of the draft of this judgment Mr Hobbs and Miss Himsworth point out that the draft judgment did not deal with uses of "Reed" other than in the copyright notice or the logos. But this was unnecessary. So far as Version 1 is concerned, it is accepted that it amounted to trade mark infringement and passing off so the details do not matter. Mr Hobbs and Miss Himsworth do not identify the other uses to which they refer. According to Mr Howe and Miss Michaels in their reply submission there is only one such instance: a three month period on Version 2 of use of the phrase "home to totaljobs the brand new online recruitment service from Reed Business Information, the leading business information publisher". The judge quoted this (para. 38) and noted that there was no evidence about it. He made no explicit finding that it infringed or amounted to passing off. Accordingly there was nothing about this use which could form a separate subject of appeal. In any event the point is trivial, the use was discontinued before the full commercial launch of the site.
The own name defence
That being so, it is not necessary to consider whether an own name defence is made out. Nonetheless, since the matter was fully argued, I shall indicate my views, both so far as passing off and infringement are concerned. They fall to be considered separately. The debate is actually rather surreal: as a practical matter the passing off own name defence is very narrow. Only if the infringement defence were even narrower – so slim as to be practically non-existent – could anything ever turn on it in a case where there is also passing off.
The own name defence to passing off
It is long and well settled that it is no defence to passing off that the defendant has or had no intention to deceive. It is also settled that there is only a very limited own name defence. It was put this way by Romer J in Rodgers v Rodgers (1924) 41 RPC 277, a passage approved by the majority of the House of Lords in Parker-Knoll v Knoll International  RPC 265:
To the proposition of law that no man is entitled to carry on his business in such a way as to represent that it is the business of another, or is in any way connected with the business of another, there is an exception, that a man is entitled to carry on his business in his own name so long as he does not do anything more than that to cause confusion with the business of another, and so long as he does it honestly. To the proposition of law that no man is entitled so to describe his goods as to represent that the goods are the goods of another, there is no exception.
Thus the English law of passing off abounds with cases where people have been prevented from using their own name. This particularly happens when a scion of some well-known family business has sought to cash in on his name at the expense of that business, as for instance Dunhill (Dunhill v Sunoptic [ FSR 426), Gucci (Guccio Gucci v Paulo Gucci  FSR 81) and Asprey (Asprey & Garrard v WRA (Guns))  FSR 31).
I have already observed that the difference between mere confusion and deception is elusive. I suppose the sort of background non-damaging confusion in this case might be the sort of thing Romer J had in mind. Once the position strays into misleading a substantial number of people (going from "I wonder if there is a connection" to "I assume there is a connection") there will be passing off, whether the use is as a business name or a trade mark on goods.
The Judge rightly observed that the passing off defence is narrow. Actually no case comes to mind in which it has succeeded. Because the test is honesty, I do not see how any man who is in fact causing deception and knows that to be so, can possibly have a defence to passing off.
Thus in this case if RBI had been causing significant deception, the fact that it was using its own name would have afforded no defence to the passing off claim, even though it had no intention to deceive. That is the admitted position as regards Version 1. But in relation to later versions there is at best no more than some minimal degree of confusion – significant deception is not shown and so, narrow though it is, I think RBI come within Romer J’s exception.
The own name defence to infringement – s.11(2)(a) = Art.6.1(a)
Clearly the words "Reed Business Information" were used as the name of a trading entity in every context in which they appeared on the totaljobs website. It matters not that the word "Ltd." or some other indication of incorporation is added. The entity was in fact Reed Business Information Ltd. So it was using its own name. So much was common ground. Two questions then arise. First can a company avail itself of the defence? And second, if it can, can the defence apply when confusion is in fact being caused?
The company name point has been considered in several cases. But there is no need to go further than Scandecor Developments v Scandecor Marketing  2 CMLR 30. For in that case, Lord Nicholls (with whom the other members of the House agreed), at para. 54 said that the "better view is that a company can claim the protection of this provision". That may be taken to be the settled opinion within the UK. Lord Nicholls went on to hold, however, that the position is not acte clair. Consequently the question of whether the defence could apply to a company was included in a reference to the ECJ. Scandecor was subsequently settled so we do not have an ECJ view on the point. It follows from all this that in this Court the defence must be held to be available to RBI, subject only to any question of a reference. I add just three points:
The argument based upon the UK Interpretation Act 1978 cannot assist – this is a European provision and the meaning cannot turn on a particular national rule of interpretation;
It would be very strange if no company could avail itself of the defence. Think, for instance, of a company formed to take over a business established under an individual’s name and having his name. It would be outrageous if the defence were lost upon incorporation.
Any fear that dishonest people might form companies with misleading names so as to take advantage of the defence is easily removed by the use of the proviso – such a deliberate attempt to avail oneself of another’s mark would not be an honest practice.
What then of the proviso? Mr Howe submitted that where a man was using his own name then, even if he was causing some confusion, the defence could apply. Otherwise it would be virtually devoid of content, only applying in Art. 5.1(a) and (b) cases where the mark proprietor is not in fact using his mark.
Mr Hobbs submitted that was indeed the case. He submitted that the phrase "honest practices in industrial and commercial matters" could not apply where the defendant was causing confusion, even if at the time he believed he was not. And he certainly could not have a defence once he knew he was causing significant confusion.
Mr Hobbs’ argument ran thus:
The phrase "honest practices ...." comes from the Directive and should be interpreted accordingly;
The 12th Recital of the Directive refers to the Paris Convention for the Protection of Industrial Property and the necessity of the Directive being "entirely consistent" with it;
Art. 10bis (2) of the Paris Convention says that:
"Any act of competition contrary to honest practices in industrial or commercial matters constitutes an act of unfair competition"
So the Paris Convention is the origin of the phrase to be construed and its construction must be the same as in the Paris Convention.
The Convention in Art. 10bis(3) goes on to particularise "honest practices etc.": saying:
The following in particular shall be prohibited:
So acts which create confusion do not fall within the meaning of "honest practices etc."
Moreover, submits Mr Hobbs, a particularly learned and knowledgeable commentator, Professor Bodenhausen, the Director of BIRPI (Bureau for the Protection of Intellectual Property, now replaced by WIPO, the World Intellectual Property Organisation) said "It is immaterial whether those acts are committed in good faith, although good faith may have an influence on the sanctions to be applied" (BIRPI Guide to the application of the Paris Convention as revised at Stockholm in 1967).
Mr Hobbs told us that certain writers even went so far as to hold the view that no Art 6.1 defence could apply if the defendant was using the mark complained of as a trade mark.
Learned though Mr Hobbs’ argument is, it limits the scope of the defence too much. First it is now clear that a trade mark use by the defendant may be within its scope. This court so held in Premier Luggage v Premier Company  FSR 5. The owner of the registered trade mark "Premier" sued in respect of, inter alia, swing tags bearing the defendant’s full name "The Premier Company (UK) Ltd.". After holding that there was no passing off or Art.5.1(a) infringement in respect of these, Chadwick LJ (with whom the other members of the court agreed) said (para. 43):
If the use by Premier UK of plc swing tags does not amount to passing off, the claim for infringement of trade mark based on that use must fail also. This is because, as the judge recognised, there was no desire on the part of Premier UK to take unfair advantage of the reputation or goodwill of Premier Luggage. The swing tags were used for a legitimate commercial purpose – to identify the trade origin of Premier UK’s product and to take advantage of the existing goodwill and trade connections of Premier Decorations. The use of the name ‘The Premier Company (UK) Limited and the address ‘Premier House’ on the swing tags is within s.11(2)(a) of the 1994 Act.
Since the oral argument before us, the ECJ has confirmed what this court held in Premier; that trade mark use can fall within the scope of the Art.6.1 defence. Gerolsteiner Brunnen v Putsch Case C-100/02 (judgment 7th January 2004) was about a conflict between the registered mark Gerri (for inter alia mineral water) and an alleged infringement "Kerry Spring" for Irish mineral water from the Kerry Spring sold by a company called Kerry Spring Water. The referring court held "Gerri" and "Kelly" confusingly similar (by reason of aural confusion) and that there would be infringement under Art.5.1(b) unless there was an Art. 6.1 defence. The questions for the ECJ were:
The Court clearly held that use as a trade mark was not "appropriate for determining the scope of Art. 6" (para. 15). It was enough that "Kerry" was of geographical significance to bring it within the scope of Art. 6, even though it also had some trade mark significance. Once an alleged infringement is within one of the heads of Art. 6 the only further question is whether it is "used in accordance with honest practices etc."
Quite apart from the Premier and Gerri/Kelly decisions, it would make no sense to exclude trade mark use from the own name defence For when a man uses his name in connection with his goods or services he is using it as a trade mark – to tell you "This comes from me – John Doe". Sometimes people think that use of a name on goods or for services is not use as a trade mark. They contrast use as a trade mark with use "just as a name". But use of a name in connection with goods or services, even in small print, is trade mark use, though not of the upfront in-your-face kind the subject of vast advertising spend.
The Court also gave guidance on the meaning of "honest practices etc.". It did not go down the route suggested to us by Mr Hobbs. On the contrary it said that the "condition of honest practices constitutes in substance the expression of a duty to act fairly in relation to legitimate interests of the trade mark owner" (para. 23). The test is for the national court to carry out an overall assessment of all the circumstances – and in particular to assess whether the defendant "might be regarded as unfairly competing with the proprietor of the trade mark" (para. 26).
It is worth noting also that the Court clearly contemplated that there must be some degree of co-existence. It said that a conclusion of aural confusion (by implication alone) is not enough to conclude that use is not in accordance with honest practices (para. 25). It recognised here that in a market of 15 going on 25 member states there is a real risk of phonetic similarities arising. One sees here a recognition that the Community needs something in the nature of a doctrine of honest concurrent use, just as nineteenth century England developed such a doctrine (see GE TM  RPC 297). There are some kinds of case where people have to put up with some degree of confusion and the public has to get used to it. Provided the parties behave fairly and reasonably it works.
Mr Hobbs’ response to the Gerri/Kelly case was to attack it. He said it was inconsistent with paragraphs 51-55, 63-64 of the BMW case, paragraph 28 of the Chiemsee case  ECR I-2799, paragraphs 51-56 of the Arsenal case  RPC 9 at p.144, the 12th Recital to the Directive, Art.10 bis of the Paris Convention, and Arts. 2(1) and 39 of the TRIPS agreement. That is to suggest that the Court has completely blundered. Not only do I not agree, but it is not for me or any national court to question the latest guidance from the Court. The fact is that trade mark law is and the concepts involved are much subtler and more complex than an outsider would, at first blush, expect. The Court is naturally evolving its views on the problems involved and the often conflicting policy questions.
I conclude from Gerri/Kelly that a man may use his own name even if there is some actual confusion with a registered trade mark. The amount of confusion which can be tolerated is a question of degree – only if objectively what he does, in all the circumstances, amounts to unfair competition, will there also be infringement. In practice there would have to be significant actual deception – mere possibilities of confusion, especially where ameliorated by other surrounding circumstances (mere aural confusion but clearly different bottles) can be within honest practices. No doubt in some cases where a man has set out to cause confusion by using his name he will be outside the defence (cf. the English passing off cases cited above) – in others he may be within it if he has taken reasonable precautions to reduce confusion. All will turn on the overall circumstances of the case.
Here RBI never set out to cause confusion. All they ever wanted to do was to associate their name with their totaljobs website. When they found that Version 1 caused some confusion they progressively reduced the use of the name Reed Business Information. At no time after Version 1 did they have any evidence of any real substantial deception of the public. I think they were acting in accordance with honest business practices thereafter.
The ECJ has not yet ruled on more subtle questions concerning "honest practices." Suppose the defendant at the time believes that what he is doing will not cause substantial deception or confusion and will not amount to unfair competition but is shown to have been wrong. There is no doubt that he must stop once he knows that. But is he liable to compensate the trade mark owner for his past use? Here there are several possible solutions, both to the common law of passing off and the law of infringement. These are:
That the test is objective and one of simple causation – if the defendant in fact caused significant deception, albeit innocently, there is no defence. He must pay for the damage he unwittingly caused. The position for the past is the same as for the future.
That the test is objective but one of reasonable forseeability – objectively would one predict that the acts concerned would cause deception? This is the same question as "would a reasonable defendant, knowing all the facts, expect confusion?" In this connection a reasonable defendant can be expected, prior to commencing use, to have searched the relevant registers (national and European) of trade marks, and if alerted to a potentially conflicting mark to have further made reasonable investigation as to whether it has been used enough to have acquired a reputation or goodwill.
That the test is purely subjective – did this defendant at the time believe that he would not cause confusion? Such a belief might be either unreasonable or reasonable, stemming for instance from an inadequate knowledge of the reputation of the other man.
Notwithstanding the fact that I espoused meaning (ii) in Cable & Wireless v BT  FSR 383 at p.391, I am inclined to favour solution (i) - that the defendant must compensate the trade mark owner if he has even unwittingly caused damage by significant, but unintended deception. I think that is what an honest man would do. Moreover the language of the provision rather suggests that the actual defendant’s state of mind is irrelevant – referring as it does to "honest practices" rather than "honest use."
I say that although I note that Prof. Bodenhausen perhaps suggested a possible defence here. He indicated a qualification about good faith possibly "having an influence on the sanctions to be applied." This is Delphic. If "sanctions" means remedies (as I suppose it must, for I can think of nothing else) then, as far as the travaux préparatoires to the Stockholm amendment of the Paris Convention are concerned, things are not quite open and shut. It would be interesting as an academic exercise to go back to Prof. Bodenhausen’s sources, said in the relevant footnote to be the travaux préparatoires for the Hague revision (1925). Mr Hobbs’ diligent researches did not take us back that far. And I am not surprised: enough is enough.
Solution (i) is also that adopted by the common law in relation to passing off so far as damages are concerned. Lord Parker put it this way in Spalding v Gamage (1915) 32 RPC 273 at p.283:
Nor need the representation be fraudulently made. It is enough that it has in fact been made, whether fraudulently or otherwise, and that damages may probably ensue, though the complete innocence of the party making it may be a reason for limiting the account of profits to the period subsequent to the date at which he becomes aware of the true facts. The representation is in fact treated as the invasion of a right giving rise at any rate to nominal damages, the inquiry being granted at the plaintiff’s risk if he might probably have suffered more than nominal damages.
In the end therefore the solution is simply that an own name defendant who has infringed and also passed off will be injuncted for the future (by which time he will have actual knowledge that he is causing deception) and will have to pay compensation both for passing off and infringement for the past use, having, albeit unwittingly, caused real damage.
Applying that to this case, if therefore the use of the copyright notice had caused passing off, i.e. amounted to a really damaging misrepresentation, I would have held that RBI should pay appropriate compensation not only for passing off but also for Art.5.1(b) infringement irrespective of its state of mind or whether objectively speaking confusion could be expected. The point is academic however; as I suspect it will always be when the own name defence is raised. Moreover, it is difficult to imagine how compensation for infringement could be any different in amount from that for passing off.
The Yahoo banner use
This use commenced with Version 2 and ran through a number of versions ending in late 2000. As anyone who uses internet searches knows, in addition to the results of a search under a particular name or phrase, one often gets unasked for "banner" advertisements. Most of the time they are nothing but an irritation and are ignored. But you can, if you wish, "click-through", i.e. click on the banner and be taken to the advertiser’s site.
Those who wish to have their banner advertisements pop-up in this unasked for fashion have to pay for it, paying the appropriate search-engine company. Banners can be totally random or tied to particular search terms. Here RBI paid Yahoo for a totaljobs banner linked to the search terms "recruitment", and "job." They were given a free extra of their own name. Yahoo chose "reed" rather than "reed elsevier". The consequence is that a Yahoo search done under any of these terms was apt to cause the totaljobs banner to appear. If it did, the user could click-through to the totaljobs website.
The Judge held that when the banner was triggered by the word "Reed" there was infringement, although loss could not be proved merely because the banner appeared. By implication he held that only if there was a click-through could there be any loss. The Judge’s finding is not explicitly tied to any particular form of infringement.
I am unable to agree with this. The banner itself referred only to totaljobs – there was no visible appearance of the word Reed at all. Whether the use as a reserved word can fairly be regarded as "use in the course of trade" or not (as to which I express no opinion), I cannot see that causing the unarguably inoffensive-in-itself banner to appear on a search under the name "Reed" or "Reed jobs" can amount to an Art.5.1(b) infringement. The web-using member of the public knows that all sorts of banners appear when he or she does a search and they are or may be triggered by something in the search. He or she also knows that searches produce fuzzy results – results with much rubbish thrown in. The idea that a search under the name Reed would make anyone think there was a trade connection between a totaljobs banner making no reference to the word "Reed" and Reed Employment is fanciful. No likelihood of confusion was established.
That is not to say, of course, that if anyone actually clicked through (and few did) and found an infringing use, there could not be infringement. Whether there was or not would depend solely on the site content, not the banner.
If this had been an Art.5.1(a) case then the position might have been different. For then there would have been no requirement to prove a likelihood of confusion. The question would appear to turn on whether the use of the word "Reed" by Yahoo at the instance of RBI properly amounted to a "use in the course of trade" as to which, as I say, I reserve my opinion. It may be that an invisible use of this sort is not use at all for the purposes of this trade mark legislation – the computers who "read" sets of letters merely "look for" patterns of 0s and 1s – there is no meaning being conveyed to anyone – no "sign."
I can conveniently deal with passing off here too. The Judge held that there was passing off by substitution. I cannot agree. Passing off by substitution is where a trader having accepted an order for brand X supplies brand Y in such circumstances that the customer is unlikely to notice the substitution and is thus misled, see e.g. Lever Bros v Mabro (1912) 29 RPC 225. That is a hundred miles from a consumer conducting a search under the name Reed and finding a banner which on its face has no connection with his search term. Again, if he clicks through and finds misleading material on the site, there could be passing off – but that would be as result of what is on the site, not the Yahoo use.
The case here was never properly identified in the pleadings and was allowed to grow without control. I think this is a pity – it became an elaborate matter involving extensive evidence largely about facts, which could not be in dispute. Since the Judge dealt with it so must we.
Search engines have elaborate indexing systems which can take account not only of visible matter in a website but also matter which never appears visibly – so called "metatags". The consequence is that whenever a user conducts a search which includes a word in a metatag, the search results will include that site along with all other sites which use that matter, either in visible or hidden form. That is why, when you conduct a search, some of the results appear to have nothing to do with your search term. Sometimes the metadata is translated into visible text in a search engine results page, for instance in a phrase "totaljobs is the new recruitment service from Reed Business Information."
The order in which search results appear is a matter for the particular search engine’s secret system. Naturally website owners play the game of trying to get their site to appear high up in any search. There was an unpleaded allegation that RBI had played such a game, with a suggestion that it was particularly aimed at getting searches for "Reed" or "Reed jobs" to appear high with some sort of dishonest motive or at least enough of an improper motive to defeat the own name defence. Without pleadings, the Judge in my view ought not to have allowed this to be gone into. In the event he did and rejected the point.
But he nonetheless held that RBI’s metatag use was passing off and an infringement. The totaljobs site, in various versions, had the words "Reed Business Information" as a metatag. Evidence was led as to what happened if a search under the phrase "Reed jobs" was made. In all cases where totaljobs was listed, it came below the Reed Employment site in the search results (which, as is usual, included many other results, irrelevant to both sides). Obviously anyone looking for Reed Employment would find them rather than totaljobs. I am unable to see how there could be passing off. No-one is likely to be misled – there is no misrepresentation. This is equally so whether the search engine itself rendered visible the metatag or not.
Nor can I see any scope for trade mark infringement under Art.5.1(b). Assuming metatag use counts as use of a trade mark, there is simply no confusion here. I confess to not following the Judge’s reasoning on the point. He said that the "ultimate purpose [of the metatag] is to use the sign to suggest a connection which does not exist." But purpose is irrelevant to trade mark infringement and causing a site to appear in a search result, without more, does not suggest any connection with anyone else. The Judge also said "the sign in a metatag is being used to suggest that this site is to be treated in a manner appropriate to the way in which the trade mark owner’s site should be treated." Even if that be so (and I doubt it, no "suggestions" are being made) it does not answer the "global assessment" test laid down by the ECJ.
Again I would wish to reserve my opinion in relation to Art.5.1(a). There are several difficult questions:
First, does metatag use count as use of a trade mark at all? In this context it must be remembered that use is important not only for infringement but also for saving a mark from non-use. In the latter context it would at least be odd that a wholly invisible use could defeat a non-use attack. Mr Hobbs suggested that metatag use should be treated in the same way as uses of a trade mark which ultimately are read by people, such as uses on a DVD. But in those cases the ultimate function of a trade mark is achieved – an indication to someone of trade origin. Uses read only by computers may not count – they never convey a message to anyone.
If metatag use does count as use, is there infringement if the marks and goods or services are identical? This is important: one way of competing with another is to use his trade mark in your metatag – so that a search for him will also produce you in the search results. Some might think this unfair – but others that this is good competition provided that no-one is misled.
If metatag use can fall within the infringement provisions of Art 5, can the defences under Art. 6.1(a) apply, for instance the own name defence? The Judge thought they could not because the use was invisible. That makes little sense – why should visibility be irrelevant to Art.5.1 (both limbs) but relevant to Art 6.1? Mr Hobbs felt unable to support the Judge in this regard, accepting (again as a member of the realist school) all defences were in principle available to metatag use. However, as I say, I do not have to consider this further.
Accordingly I hold that there was no infringement by the use of the "Reed Business Information" metatags.
Other invisible uses
There is a directory of services called DMOZ, compiled by human volunteers. totaljobs.com was included within it. A search under the phrase "Reed jobs" produced it as the third of three results. The first was Reed Employment and the second a business somewhere in Connecticut called "Paxton-Reed Recruitment". The totaljobs site was described as "A searchable database of over 12,500 live job opportunities, including IT permanent, IT contract and cool web jobs from Reed Business Information."
I am unable to see how this alone could cause any confusion. Hence there is no passing off or Art.5.1(b) infringement. Actually the DMOZ allegation was not pleaded and the Judge made no explicit finding of infringement or passing off by it..
There were some other minor invisible uses involving sites called lawmoves and coolwebjobs. The Judge did not rule on them separately.. Before us, the parties, under a little pressure, agreed what should be done, namely a short recital in the order. So I say no more.
Use of "Reed Business Information" on invoices, press releases and promotional materials
The Judge rejected the infringement claim relating to these. Reed Employment cross-appeal. There is little that needs to be said. The invoices were all to people who knew exactly who they were dealing with – people who had placed advertisements on the totaljobs site. No-one was or could be confused. One only wonders why it was thought worthwhile to raise the point at all. The Judge was right to reject it as being an infringement at all and going on to hold, as he did, that in any event it fell within Art.6.1(a).
The press releases were essentially about totaljobs.com itself. Incidentally they included information of this sort:
Notes to Editors
Pumfrey J quotes similar material from another press release at paragraph 78.
I am quite unable to see how this could cause any confusion or deception. It is all true and honest. Those who already had vaguely wondered whether there was any connection between Reed Elsevier and Reed Employment might have been disabused – they certainly would not be caused even to wonder by this sort of stuff. Again one wonders whether a sense of commercial reality is missing to find this point raised at all, and even more so on appeal.
The promotional material (essentially that handed out at the Harrogate trade fair in 1999) was issued to RBI’s clients – who on the Judge’s findings would not be confused. It stands in the same case.
For passing off there has to be a misrepresentation causing damage. I have above held that save in respect of the unappealed findings concerning Version 1 and the logos, there was no misrepresentation. It follows that save in respect of the unappealed matters, the claim in passing off fails. The Judge also thought that the decision in passing off followed the infringement issues – to the extent that he was working on Art.5.1(b) infringement I agree.
The inquiry as to damages
The second judgment was concerned with the form of order following the first judgment. A mass of evidence was put in and the hearing lasted three days followed by a reserved judgment. I have to say that I think this was unsatisfactory. At issue was whether or not an injunction or merely a declaration should be granted, whether there should be an inquiry as to damages, and costs. But more was gone into here – the inquiry as to damages seems to have not only started but got well under way without an order for it or pleadings defining the issues.
Pumfrey J held that damages could be awarded on a "user" basis, saying:
Any case in which the claimant cannot show lost sales or loss of business must be approached on the ‘user’ basis: how much should the defendant have to pay for the use he has made of the sign?
He held that there should be an inquiry for the visible uses which were found to be infringements and passing off, for the Yahoo use (which he held was small but not entirely negligible) and for the invisible uses. In relation to the latter it was to be on the basis of a reasonable royalty as between willing licensor and licensee – and on the footing that click-through would be vanishingly small. Quite why anyone would pay anything to appear in search results where they got no click-through was not explained.
Mr Howe contends that none of this should have happened, that damages should have been assessed on a summary basis since they are obviously small. The result of this appeal is to reduce considerably the ambit of any inquiry – so his contention gains force.
Prior to the CPR the position where a trade mark or passing off claim succeeded at trial was that the successful claimant would be entitled to an inquiry as to damages unless the court was satisfied it would be fruitless, McDonalds v Burger King  FSR 112. If the court was in doubt, the inquiry should be ordered, though at the claimant’s risk as to costs.
Mr Howe submitted that the CPR enabled the court to take a more adventurous course – in a case where the damages were small it should boldly apply the overriding objective and assess them summarily.
I have some sympathy with Mr Howe’s submission, but do not accept it, if, by summary assessment, he means the court should just pluck some small sum out of the air. On what basis could the court work? However I do think that the court, where it thinks the damages are likely to be negligible or small, can use its case management powers to stop things getting out of hand. It can, for instance, require the claimant to put in a statement of case together with supporting evidence before requiring the defendant to do anything. The cost in money and time of that is likely to cause a claimant who in reality has little to gain to think twice. Again the court can order the trial of quantum to be on paper only unless a case for cross-examination is made out. Disclosure can be restricted or even done away with. A time limit for the hearing can be imposed. All these things are ways of case-managing the problem. It is not for the Court of Appeal to deal with this. It is a matter for the Judge, to whom the inquiry must be remitted in the light of this Court’s findings.
I would add only one further matter in relation to damages. I am by no means convinced that the "user" principle automatically applies in trade mark or passing off cases, especially where the "mark" concerned is not the sort of mark available for hire. The ordinary case is one that just protects goodwill. For damages to be awarded on the user principle is close to saying there is no damage so some will be invented. It is not the same sort of thing as having to pay for use of an invention (the basis of the user principle in patents). At present there is no clear finding that the inquiry should proceed on a user basis. Whether it should do so will be a matter for the Judge – who thus far has said no more than that he "apprehends" that a user basis will be used.
The Judge decided to deal with costs to date and did so. He ruled that an open offer of settlement made on 13th November 2002 been beaten, that the claimants had run up some costs unnecessarily and in the end awarded them 80% of the costs of the trial on liability. As regards the costs of the form of order dispute, he ordered the claimants to pay some of the costs (on a dispute they started without substance) and reserved the remainder to the end of the inquiry. Mr Howe says the Judge, in the circumstances of this case, ought to have waited until the conclusion of the trial of quantum, before dealing with any of the costs. The Judge had already formed the view (or at least had the "impression") that "the expenditure of time and effort at trial was disproportionate to the results actually achieved." If, as Mr Howe contended, the award of damages were to be also trivial, the court could take an overall view on costs against the result achieved, and apply the principle in Home Office v Lownds  1 WLR 2450. If that were done at the end of the day, the costs expended (we were told £1.155m for the claimants and £1m for the defendants just down to the end of the liability trial) would appear disproportionate to the results.
Since we take a different view from that of the Judge on the matters raised in this appeal, the submission falls away and we will have to hear fresh submissions on costs. Those submissions ought to deal with two matters raised before us during the course of argument. First, by the time the letter before action was sent, RBI had decided to reduce the amount of use of Reed considerably, yet so far as we can see, Reed Employment were not told that and the action was allowed to start essentially on a false basis. Second, RBI refused a number of offers to go to mediation. Mr Howe told us there were serious settlement discussions, but they are not the same as mediation. A good and tough mediator can bring about a sense of commercial reality to both sides which their own lawyers, however good, may not be able to convey. In this connection the decision of Lightman J in Hurst v Leeming  1 Lloyds Rep. 379 may be in point.
In the result I would allow the appeal. If my brothers agree I would make the following directions as regards argument on the form of order:
RBI to serve its proposed form of order within 7 days;
Reed Employment serve its proposed form of order within 7 days thereafter, doing so by the use of strike-out and italics on RBI’s draft;
Skeleton arguments from both sides within 7 days thereafter.
Whether a hearing would be necessary should be resolved thereafter. There is of course no reason why the parties should not also enter into sensible discussions now – assisted if both sides thought it a good idea, by a mediator even at this late stage.
Lord Justice Rix
Lord Justice Auld
I also agree.
Martin Howe QC and Amanda Michaels (instructed by Olswang) for the Appellants/Defendants.
Geoffrey Hobbs QC and Emma Himsworth (instructed by Slaughter & May) for the Respondents/Claimants.
all rights reserved