On March 15th the IP Conversazione took place in Oxford as part of the 17th Annual Oxford International Intellectual Property Law Moot. The moot pitted students of IP law from around the world against each other to argue a trade mark case involving two identical trade marks used for different but reasonably similar goods. Roger Teichman and I were invited to speak to participants about work related to this question. This post summarises some of my comments and adds some details that have emerged since.
Philosophy of Trademarks
Roger drew our attention to Borges’ classification of animals that includes headings such as: “embalmed ones” and “others”. This classification, though odd, might well serve a purpose and it can be used to group animals together that are similar with respect to the classification. Roger used the example to illustrate a point: a judgement that animals or goods are similar requires a reference point from which similarity is derived. Without the reference there is no similarity.
This goes to the heart of an enduring problem for lawyers managing a trade mark register: which marks are similar? With trademark law the real reference point is the customer who might be deceived about the origin of the product they are buying. The classification of goods and services that underpins the trade mark register should be a summary of what is considered similar from this perspective. But this isn’t what happens, partly because of the register’s historical origins and partly because of the way that firms and trade mark offices currently use the register.
Consider an example: will the customers of Sky UK believe that fleet management systems marketed by Joao Luis Martins Esteves LDA from Portugal under the name Skyview are products of Sky UK, should they ever encounter them? Will they view the likely quality of the fleet management system in light of their experience of Sky UK? It seems unlikely, yet the structure of the current trade mark register led a panel at the European IP Office to reject an application for an EU wide trademark for Skyview based on opposition by Sky UK. The reason given was that the marks are similar visually, aurally and that the goods and services protected by these two marks are similar. This is where one might depart from the panel’s judgement, knowing what Sky UK actually sell. However, if the reference point is the trade mark register then the conclusion that signalling apparatus and instruments are similar to electronic tracking systems may not seem too far fetched. Sky UK included signalling apparatus in the list of goods and services for which their mark was registered and that was enough to win this case.
History of Classifying Trademarks
Trademark registers around the world have their origins in the registers created in Britain, the United States and Germany between 1874 and 1894. These were created in response to demand by businesses who came to rely on their brands in the course of trade over increasingly large territories and were looking for protection of these brands against copying by rival firms. The original register in Britain was divided into a set of 50 classes of goods, while the first centralised German register had 42 classes of goods. German law also required firms to list the goods their marks were registered for. While the British register did include some examples within each class there was initially no comprehensive list of goods in either country. The centralised German register set up in 1894 was managed by trade mark divisions specialising on specific classes of goods and services which had the task to rule on similarity of marks. By the 1930’s several hundred decisions on similarity of goods and services had been made by the German office and these were assembled into a publication structured using an alphabetical list of goods. In 1929 an international committee of experts agreed a classification of 34 classes and used lists furnished by the British and German patent offices to generate an alphabetical index of goods linked to these classes. This classification is largely still with us today, although it has been extended and amended and now consists of 45 classes. Viewed on the basis of the goods and services on offer in 2019, the classification seems almost as arcane as the one proposed by Borges for animals: on the EU register class 9 (the second largest class) contains more marks than the smallest 13 classes together; it is also the class in which Sky and Skyview are registered. This lopsidedness of the class structure means that it is more likely that some products classified by this system will be found to be similar than might be the case if the classes were more evenly filled.
It is hard to see the classification being reformed soon. It is now used around the world and is the basis for many business decisions and legal rulings, such as the one cited above. That decision illustrates the importance of the classification – it is often used as a short cut to determine similarity of the goods or services being brought to market. Now a further consideration enters the frame: how often do firms restrict their applications to goods and services they really are selling? At least if this were the case, the historical oddities in the structure of the class list might not matter too much.
It turns out that many applicants will seek to register their marks more broadly than is justified from the perspective of their business activities. It also turns out that applicants tend to use widely used words when constructing their brand names: this fact is wonderfully illustrated by the recent work of Beebe and Fromer who analysed use English language terms in brands registered in the US.
The first of these behaviours contributes to clutter: the trademark register contains many entries that do not reflect actual usage of the mark, making any judgements on similarity and potential for confusion unreliable: if Sky UK’s marks aren’t being used for signalling apparatus in any market related to fleet management nobody is going to get confused when Skyview is added to the register. But many applicants get away with not just registering overly broad specifications for their marks, but also then using those specifications to limit or prevent applications made by others. It is possible that the CJEU will restrict this type of behaviour in the near future when it rules on Sky v Skykick , a case from the UK in which the defendant has argued that Sky UK made its application in bad faith and should therefore lose it.
Following the approach taken by Beebe and Fromer, consider how often certain marks appear on the EU register. First, consider the most frequent signs which consist of the letters in the alphabet, and are usually the basis for a figurative mark. The figure below provides the frequency with which letters appear on the register, as well as the frequency of the number 1.
Now consider the most frequent words in the EU trademark register used by multiple applicants. This list overlaps with the top 20 list for the US (bold face words) to some degree.
Is it a surprise to find the word Sky at number 1 on this list? The table below shows the top 20 words in the European register and the number of disputes (Clash) around these words at the European trademark office.
Notice also, that the number one makes another appearance. Disputes are particularly frequent for words that are short, such as ECO or ONE.
All in all the fact that we rely on a classification that is somewhat dated and the fact that users like to cluster around the same words when constructing marks means that the pressure to ensure the register really reflects usage of marks by firms is increased. While applicants who did not think of an effective brand name first, cannot expect to be able to use such a brand, applicants who did move first also cannot expect to hog the brand in markets which they don’t operate in. If firms can regularly get away with overly broad specifications (goods and services lists) of their marks and if they can make use of these specifications to limit rivals’ usage of effective marks, then this undermines the point of including goods and services descriptions in the classification. They can no longer be relied on to indicate that marks are being used in ways that will confuse consumers, since consumers are unlikely to encounter the marks in a confusing manner.
Here is one example from Europe, illustrating that many marks there have been registered more broadly than is warranted and that these marks have been used to limit the registration of marks by other firms for goods and services for which the first mark has not been used. This example focuses on marks first registered between March of 2005 and March of 2007. At the time of registration firms faced the same costs for registering in one class as they faced for registering in three: many firms went for three classes. A decade later when these marks came up for renewal, the EUIPO had changed the fees for renewal: now marginal costs of additional classes were positive. The fees changed end of March 2016. Many firms renewing after this date dropped classes they did not need from their marks. The two figures below illustrate which classes were dropped for marks renewing before March of 2016 (top) and after this date (bottom).
Clearly many more marks were narrowed down in response to the change in fees. The interesting point is that opposition cases brought on the basis of these marks between 2005 and 2016 were sometimes brought against marks registered in those classes that were removed from the opposing mark at the end of the decade. It seems likely that the opposed applications might have succeeded, had they challenged the opposing mark on the basis of non-use, but this rarely happened in this period.
The figure below shows the number of opposition cases in various classes, based on marks filed in the year before (hollow symbol) and after March of 2006 and the share of those cases that rested on overlap between opposed and opposing mark in a class that was later removed from the opposing mark. While the shares are small, the number of such cases is in the hundreds in the most populated classes. This example shows that overly broad classifications have real effects for applicants.
The question we are currently working on is how big the problem of excessive breadth in trade mark specifications is in Europe and what type of costs this creates for other users of the register. Any feedback on these questions from trade mark users or attorneys would be much appreciated.