Section C: Did examples 8 to 11 involve infringement? (Ground 2)
45 There is no dispute that Examples 8 to 11 are substantially identical with or deceptively similar to the trade marks in suit. Thus, for Hells Angels Australia to succeed at trial it was, in principle, only necessary for it to establish for the purposes of s 120(1) that Redbubble had used each of Examples 8 to 11 as a trade mark. It proved this by demonstrating that the trade marks officer of Hells Angels Australia, Mr Hansen, had viewed each of Examples 8 to 11. It did not succeed in demonstrating that any other person had ever viewed them in Australia and the trial judge accepted that it was in fact only Mr Hansen who had done so. Even though Mr Hansen viewed the images depicted in Examples 8 to 11, he did not place any orders for merchandise bearing these images although, as already noted, he did take a screenshot of each. As we have already observed, the servers on which Redbubble's website is hosted are located in the United States. It may be accepted that although directed to the world at large rather than any specific nation, the website is intended to be accessed by persons in Australia and, indeed, for persons in Australia to purchase merchandise through the website.
46 At trial, Redbubble did not submit that the trial judge should have concluded from the fact that the website was hosted from servers located in the United States that there could be no trade mark infringement in Australia if no one in Australia, apart from Mr Hansen, had ever viewed the images depicted in Examples 8 to 11. Even so, it sought on appeal to advance this proposition in ground 1 of its Notice of Appeal. Leave was necessary for this course to be pursued but Redbubble declined to seek that leave. During the hearing of the appeal Ground 1 was abandoned. It is nevertheless useful, however, to say something about the point although we need not reach any conclusions about it.
47 The judgment under appeal is the second proceeding in which Hells Angels Australia has successfully sued Redbubble for trade mark infringement, with the first resulting in the 2019 judgment. In the first trial, the trial judge found at [469] that '[t]he capacity to engage, in Australia, through the website, as Mr Hansen did, constitutes use in Australia by Redbubble' (our emphasis). It is not clear from the record of the first trial whether the images involved in that case had been downloaded by persons in Australia other than Mr Hansen. But in the case of Examples 8 to 11 the trial judge found that the images had not been viewed in Australia by anyone apart from Mr Hansen.
48 For the reasons which follow, to the extent that [469] of the 2019 judgment deals with the situation disclosed by Examples 8 to 11, we would reserve the correctness of that statement for a case where it is directly raised.
49 There are at least three questions which would need to be determined if that issue were to be raised.
50 First, it would be necessary to examine the legal mechanism which explains why a trap viewing (i.e., that which precedes a trap purchase) does not itself directly involve an infringement. In Ward Group Pty Ltd v Brodie & Stone Plc [2005] FCA 471; 143 FCR 479 ('Ward') Merkel J relied at [54] on the former s 123(1) (trade mark applied to goods by or with consent of owner) to reach the conclusion that a trap purchase did not in itself involve an infringement, but that provision has since been repealed and replaced and can no longer serve the purpose to which Merkel J put it. It is likely that the new s 122A (exhaustion of trade mark rights), which replaced the former s 123(1), cannot be pressed into service to achieve that purpose either, although we would at this stage express no concluded view about this.
51 It is perhaps more likely, contrary to the views of Merkel J in Ward at [46], that the provision of images to a trap viewer by way of download from a website will constitute an authorised use for the purposes of s 8(1) ('A person is an authorised user of a trade mark if the person uses the trade mark in relation to goods or services under the control of the owner of the trade mark' (emphasis in original)). If so, then some of Redbubble's actions in providing to Mr Hansen the images in Examples 8 to 11 may have been an authorised use within s 8(1) since it was Mr Hansen who was in control of the application of the images to the virtual merchandise displayed on the website. If that were so, Redbubble would have been entitled to use the images under s 26(1)(a) ('the authorised user may use the trade mark in relation to the goods and/or services in respect of which the trade mark is registered…') and therefore Redbubble would have been authorised by the Act to do most of what it did in the case of Mr Hansen. If authorised by the Act, s 122(1)(e) means that the provision of the images to Mr Hansen cannot have involved an infringement for there will be no infringement where 'the person exercises a right to use a trade mark given to the person under this Act'.
52 However, these matters have not been argued. The question of whether a trap viewing of an overseas website can in itself constitute an infringement therefore need not be determined. As it happens, any assessment of this proposition is likely to be highly dependent on the particular facts. We emphasised above that these arguments may apply to some of Mr Hansen's activities, but not all. For example, the website includes a display function called a 'carousel'. On each design's individual webpage, several carousels appear, displaying designs which are related in some way to the design at hand. The user can then browse and access those other designs directly from that webpage. In this case, Examples 9 and 11 were suggested to Mr Hansen by the website's carousel function following his discovery of Examples 8 and 10. Whilst one may perhaps see the viewing of Examples 8 and 10 as having taken place under the control of Mr Hansen, it may be rather hard to see the initial and unprompted supply to Mr Hansen of the images in Examples 9 and 11 as having that quality.
53 We would note that Hells Angels' pleading is likely sufficiently broad to catch the appearances of Examples 9 and 11 on the carousel and its submissions at trial did draw a distinction between the images which Mr Hansen found by himself and those which were suggested to him by the carousel.
54 The second issue which would need to be assessed would be the geographical reach of s 120(1). This is largely uncontroversial. Section 120(1) does not have an extra-territorial operation and the use of a trade mark to which it refers is to be understood as a reference to the use of a trade mark in Australia. A statute is presumed not to have an extra-territorial operation unless its terms show that the legislature intended it to do so: Jumbunna Coal Mine NL v Victorian Coal Miners Association (1908) 6 CLR 309 at 363 per O'Connor J. That principle has been applied to the Act: Playgro Pty Ltd v Playgo Art & Craft Manufactory Ltd [2016] FCA 280; 117 IPR 489 ('Playgro') at [121] per Moshinsky J; Taylor v Killer Queen, LLC (No 5) [2023] FCA 364; 172 IPR 1 ('Taylor v Killer Queen') at [278] per Markovic J; PKT Technologies Pty Ltd (formerly known as Fairlight.Au Pty Ltd) v Peter Vogel Instruments Pty Ltd [2018] FCA 1587; 135 IPR 463 at [137] per Nicholas J.
55 The Act does not evince such an intention. To the contrary, s 4 explicitly identifies the external territories to which the Act is to apply and, significantly, omits reference to the Australian Antarctic Territory, the Territory of Heard Island and McDonald Islands, the Territory of Ashmore and Cartier Islands and the Coral Sea Islands Territory. The extra-territorial application of the Act is inconsistent with the excision of these external territories from the Act's territorial reach.
56 Section 120(1) does not therefore apply to the use of a trade mark which takes place outside the territorial confines of Australia.
57 The third issue is whether there could be trade mark use in Australia if the images involved have never been viewed in Australia by anyone apart from a trap viewer (assuming here that a trap viewing cannot itself constitute an infringement). It has been accepted that a website directed at Australia will involve trade mark use in Australia where an image is downloaded in Australia: Ward at [43] (in obiter); Christian v Société Des Produits Nestlé SA (No 2) [2015] FCAFC 153; 115 IPR 421 ('Christian') at [78] per Bennett, Katzmann and Davies JJ; Taylor v Killer Queen at [283] per Markovic J; International Hair Cosmetics Group Pty Ltd v International Hair Cosmetics Ltd [2011] FCA 339; 218 FCR 398 at [47] per Logan J; Playgro at [141] per Moshinsky J.
58 In Christian the Full Court found at [78]-[79] that there had been trade mark use in Australia in relation to the first A-SASHI mark but this conclusion was based on trap viewings and it may be that the Court in that case overlooked the fact that the use it identified may not have constituted an infringement. Indeed, at the time of Christian the former version of s 123 remained in place and the reasoning of Merkel J in Ward supported the idea that a trap viewing could not constitute use of a mark in Australia.
59 More recently, the Full Court (comprised of Yates, Stewart and Rofe JJ) in Energy Beverages LLC v Cantarella Bros Pty Ltd [2023] FCAFC 44; 170 IPR 281 ('Energy Beverages') dealt with a case where a commercial using a trade mark was available on Facebook and YouTube. It declined to infer that the commercial had been viewed in Australia during the non-use period merely because an emoji and a comment had been posted to those pages. The Court at [76] cited Ward and Christian for the proposition that '[u]nder existing authority, which has not been challenged in the present application, the mere uploading of trade mark content on a website outside Australia is not sufficient to constitute use of the trade mark in Australia'. Consequently, it concluded that the attempt to establish use of the trade mark during the non-use period failed.
60 Energy Beverages does not address the situation which arises in this case. As we shortly explain, Redbubble, in the ordinary course of its business, used its website to offer for sale in Australia goods on which it also offered, if desired, to print the images depicted in Examples 8 to 11. The present challenge is how to reconcile that proposition which appears self-evidently to involve trade mark use in Australia with the geographical limitations inherent in s 120(1). In a bricks and mortar store situated in Australia, an argument that the trader had not offered goods for sale in the ordinary course of its business because no consumers ever actually looked at the shelves where the goods were located would be hopeless. But this is not self-evidently so where one is dealing with an overseas website directed to the world at large (including Australia) which is never actually viewed in Australia.
61 There can be trade mark use in Australia by an overseas manufacturer who 'projects into the course of trade in this country' goods bearing the relevant mark: Estex Clothing Manufacturers Pty Ltd v Ellis & Goldstein Ltd (1967) 116 CLR 254 ('Estex') at 271. Estex was an application to remove the EASTEX trade mark from the register on the grounds of non-use. The High Court held that an overseas manufacturer of women's clothing who sold its goods overseas to retailers for re-sale, including Australian retailers, used its trade mark in Australia. It rejected the notion that the manufacturer had to sell the goods in Australia and it was in that context that the above statement that it is sufficient if the trader 'projects' the goods in the course of trade in Australia was made. That principle has been applied more than once but, as in Estex, in each case there was a threshold requirement that there be an actual trade or offer to trade in the goods in Australia: Moorgate Tobacco Co Limited v Philip Morris Limited (No 2) (1984) 156 CLR 414 at 433-4 per Deane J; E & J Gallo Winery v Lion Nathan Australia Pty Ltd [2010] HCA 15; 241 CLR 144 at [52] per French CJ, Gummow, Crennan and Bell JJ; Accor Australia & New Zealand Hospitality Pty Ltd v Liv Pty Ltd [2017] FCAFC 56; 124 IPR 264 at [167] per Greenwood, Besanko and Katzmann JJ.
62 The principle has also been applied in an infringement context by Moshinsky J in Playgro at [139]. However, in that case, as in Estex, there was an Australian trade in the goods, even though they had been sold - and title passed - overseas. The question of whether mere overseas projection without a local trade in the goods can amount to trade mark use in Australia is, in our view, a question of considerable difficulty. It is made potentially more complex in this case because although the website is hosted from servers in the United States there is no doubt that Redbubble conducts business in Australia. The difficulty is that that business does not appear to have involved, in the case of Examples 8 to 11, any more than projection into the Australian market without any consequent trade in goods bearing the marks. An important question is whether the necessary geographical nexus for use of a trade mark in Australia can be established by the mere fact that the trader is engaged in trade in Australia albeit not in relation to the infringing trade mark. The answers to these questions are not self-evident.
63 On the current state of the authorities, we regard the matter as undetermined. At no point in either of the trial judge's decisions does his Honour traverse these issues. We would therefore not read [469] of the 2019 judgment as resolving them.
64 As finally pressed in this Court, Redbubble's submission about Examples 8 to 11 was confined to the question of whether the trial judge was correct to conclude that the images were available to consumers in the ordinary course of trade. For the reasons we have just given, this does not include any consideration of whether the ordinary course of that trade had an Australian element to it. Thus in the way this appeal has been presented, the answer to this difficult question has been assumed to be that an overseas website projecting to the world at large (including Australia) is sufficient and the debate has been, rather, whether the projection which occurred is in the ordinary course of Redbubble's business.
65 At trial Redbubble contended that the difficulties involved in locating the images depicted in Examples 8 to 11 and the brevity of the period during which they were actually visible on the website entailed that no ordinary consumer would ever have found them. In making this submission Redbubble was tracking a similar and successful submission made to Merkel J in Ward. That case concerned the sale of the Restoria hair product, a product designed to conceal the ravages of age in men. The owner of the Restoria trade marks in the United Kingdom was different to the owner of the marks in Australia. Products bearing the UK Restoria trade marks were available for purchase on UK websites directed to the world at large. The owner of the Australian trade marks caused trap purchases of the product to be made through those websites and imported into Australia.
66 As in this case, the applicant sought to demonstrate that the trap purchases were evidence from which inferences could be drawn about what the website proprietors did in the ordinary course of their trade. This was an orthodox use of evidence about trap purchases: CC Wakefield & Co Ld v Purser (1934) 51 RPC 167 at 171 per Farwell J. But as that case shows, the circumstances of a trap purchase can sometimes make it inappropriate to draw such an inference. It was that principle that the Court applied in Ward. It declined to draw the inference sought by the plaintiff because it accepted two evidentiary propositions advanced by the respondents. The first of these was that it was unlikely that Australian consumers would seek to purchase the UK Restoria products because the Australian version was readily available for purchase in a large number of retail outlets in Australia and on the applicant's Australian websites. The second was that the UK product was more expensive than the Australian product and its purchase from the UK website would involve additional postage costs. The thrust of the defendant's contention was that these economic matters meant that Australian purchases from the UK websites were very unlikely to occur.
67 Thus, whilst the Court accepted that there had been trap purchases from the UK websites it did not accept that one could conclude from these purchases that the products were being sold in Australia in the ordinary course of the various website proprietors' businesses. As Merkel J put it at [28], there was 'no reason to expect that sales in Australia were likely to occur'.
68 Redbubble's current submission is similar in its thinking and design to this finding. It submits that whilst Mr Hansen did view Examples 8 to 11 on its website, the circumstances of his doing so were not such as to provide a proper basis for an inference that Redbubble was making the images depicted in Examples 8 to 11 available to consumers in the ordinary course of its business.
69 The trial judge was not disposed to accept this as a matter of fact. His Honour accepted that the evidence established that Examples 8 to 11 were initially identified by a director of Hells Angels Australia, Mr Nelms, who 'spent some time over a period of approximately seven hours (with various breaks throughout totalling 4.5 hours)' on the website looking for offending images which he ultimately found in the form of the images at Examples 8 to 11: [174]. Mr Nelms was situated in Slovenia when he did this. His endeavours in this regard had involved him undertaking a degree of 'inquisitive forensic examination' of the website: [178]. Following Mr Nelms's examination of the website, he shared with Mr Hansen (who was in Australia) the fruits of his labours and thoughtfully provided him with the direct links to particular webpages displaying the offending images: [178]. Mr Nelms then suggested that Mr Hansen should search for them himself to see if they were accessible in Australia from the website: [174]. This would appear to reflect an understanding on Mr Nelms's part that viewing the images in Slovenia could not advance its trade mark infringement case in Australia. On the other hand, Mr Nelms's wife, Ms Lyhne (who was also in Slovenia), seems to have confronted the same problem as her husband by seeking to have goods ordered from Slovenia sent to Australia (although Ms Lyhne appears to have limited her activity to Exhibits 1 to 7).
70 The trial judge thought that it was likely that Mr Hansen went to the links provided by Mr Nelms and clicked upon them, which then exposed the lode uncovered in Slovenia. Following this early success, and presumably emboldened thereby, Mr Hansen then probed the website further to identify the field of products to which the images might be capable of affixation: [178]. Next, his Honour accepted that Mr Hansen's successful location of Examples 8 to 11 on the website had involved industry and, in particular, had involved him searching for images using the search term 'Hells Angels' combined with a number of different search filters provided by Redbubble; specifically, its 'Newest' and 'Most recent' filters (which, for the uninitiated, took him to a set of results showing the newest or most recently uploaded images responsive to the search term): [174]-[176], [181]. The trial judge accepted that a small percentage of persons (perhaps 1.8%) conducting searches of the website use the filter 'Newest': [175].
71 These matters led Redbubble to submit that the combined activities of Mr Nelms and Mr Hansen could not be said to be behaviours exhibited by ordinary consumers. Despite that submission, the trial judge concluded that 'a consumer relevantly engaged by the force of seeking to find images so connected to his or her area of interest could have found the images, admittedly with some effort': [181]. Thus he did not accept that this 'degree of industry' ([183]) or the fact that 'the images were on the website for a short time' ([182]) prevented the conclusion that Redbubble was offering to affix the marks depicted in Examples 8 to 11 to a wide range of goods: [182]. Although his Honour did not use, in this part of his reasons, the terminology of a trader's ordinary course of trade, it is evident this is what he had in mind.
72 On appeal, Redbubble challenged this factual finding. It submits that Mr Hansen was tipped off by Mr Nelms and what Mr Nelms had done was not ordinary consumer behaviour. It relied on evidence arising from the cross-examination of Redbubble's Senior Vice President, Technology Mr Vydra and Assistant General Counsel Mr Toy to the effect that people searching on behalf of holders of intellectual property rights do not behave like ordinary consumers on websites like the one in question. Redbubble also cited statistics provided by Mr Vydra that, between September 2020 and September 2021, consumers on the website spent on average a mere three minutes and 46 seconds and viewed an average of 4.38 pages per session. This was to be contrasted to the 2.5 hours Mr Nelms had lavished on the website in Slovenia.
73 In our view, evidence of this kind does not undermine the trial judge's conclusion that a sufficiently interested member of the public could have located the images. One may leave aside any consideration of the appropriate standard of appellate review to be applied to a finding of this kind because we are not persuaded that his Honour was incorrect.
74 The four images identified by Mr Nelms were located using the term 'Hells Angels' paired with the 'Newest' search filter. Redbubble relied on the statistic that 1.8% of all searches (filtered and unfiltered) conducted on the website between September 2020 and September 2021 were sorted using the 'Newest' filter. Those same statistics show that, during that time, users who chose to search the website using a filter applied the 'Newest' filter 12% of the time. The majority of searches were conducted without any filtering option, and the most common filter that was applied was 'Most Relevant', which was used in 3.5% of the total searches and 23% of the filtered searches.
75 Based on this evidence, Redbubble submitted that 'the chances of an ordinary consumer being both interested in searching for "Hells Angels" and choosing to apply that filter must be vanishingly small'. However, it is implicit in such a submission that the chances were not zero. Further, this evidence does not demonstrate that Examples 8 to 11 were not able to be accessed by a consumer in Australia. To the contrary, it shows that they were.
76 In oral submissions, senior counsel for Redbubble submitted that the website only displayed eight images at a time. That it had this feature was not the subject of evidence at trial and, when pressed by this Court about this, it was submitted that this is an 'inherent' feature of the website which could be inferred from the screenshots of example search results from the website. It was submitted that, because of this feature, a consumer would need to scroll through thousands of results to review all of the images displayed when they searched 'Hells Angels' and that this would not realistically occur.
77 We do not think that it would be appropriate to assume that the website only displayed eight images at a time in response to searches in circumstances where the only evidence in support of this is constituted by the screenshots of various search results, including those displayed in response to a search for 'elephant'. Too little is known about the feature which resulted in the appearance of each set of eight images in the screenshots to provide a proper foundation for such a conclusion.
78 Further, we are not prepared to assume that a search conducted by a consumer in Australia in August 2021 would not have shown Example 8 (for example) as one of the first results which appeared. Such an assumption would require evidence to be adduced about the search algorithm which was used by the website at the time. There was no such evidence. Thus, we cannot accept the proposition that a consumer in Australia in August 2021 would have needed to scroll through thousands of results before arriving at any of the relevant images.
79 Indeed, based on the evidence of Mr Hansen, Examples 9 and 11 were identified with almost no effort because they were suggested to Mr Hansen by virtue of one of Redbubble's carousel functions when he was accessing the webpages for Examples 8 and 10. One such carousel function, the 'Similar Designs Carousel', displays designs that have similar titles and/or tags to the design the subject of that particular webpage. Another, the 'Artist's Carousel', displays other designs for sale by the same creator. Given that Examples 9 and 11 were uploaded by a different uploader account but shared similar tags to Examples 8 and 10, it is likely that these images were displayed under the 'Similar Designs Carousel'. As such, there is every likelihood that a consumer searching for images with similar tags (which it might be noted included the quotidian tag 'motorclub') would have also identified these images with perhaps not much effort.
80 There are therefore factual problems with Redbubble's submissions about this. However, there is a more general objection which may be raised against it. Mr Hansen and Mr Nelms used search tools provided by Redbubble. The purpose of those tools was to assist consumers to search Redbubble's large catalogue of artworks for designs which were of interest to them. The provision of those tools and that catalogue were a central component, indeed perhaps the central component, of Redbubble's business model. It sits uncomfortably in its mouth now to argue that because the search tools it provided could not easily be used to find these particular images that somehow that took them out of the ordinary course of its trade. The reason Redbubble permitted images to be uploaded into its catalogue was for the precise purpose of being available to be searched for and located.
81 In that regard one must distinguish between, on the one hand, the difficulties a consumer may encounter in finding what they are searching for on a website explicitly designed for the purpose of helping them do so and, on the other, the idea that such difficulties entail that the website is not engaged in the ordinary course of its trade. We do not think that the fact that it might be difficult to locate goods bearing infringing trade marks in a poorly laid out store can mean that the goods are not being offered for sale in the ordinary course of the trader's business. We do not think any different principle applies to a website of the present kind.
82 In short, the evidence relied upon by Redbubble in support of its present contention does not detract from two key facts, namely: Examples 8 to 11 were able to accessed by a consumer in Australia in August 2021 and a sufficiently motivated consumer could have accessed them. The primary judge was therefore correct to conclude that the degree of industry which might be required to reveal the images was no answer to infringement.
83 We therefore detect no error in the trial judge's conclusions about this aspect of the matter.