The fourth UK Phenomenology Initiative Workshop and the second devoted to HERA Physics took place in St John's College, Durham between 18 and 23 September 1995. Since the first HERA Workshop at Durham in March 1993 [1] a number of surprising and profound results have been established at HERA and many data in many areas have been published. In deep inelastic scattering (DIS) the rise of F₂ as Bjorken- x decreases, first seen in the data of the H1 experiment unveiled at Durham in 1993, has been confirmed over a wide range of x and Q². At least superficially, the HERA data appear to be consistent with the next-to-leading order approximation to the QCD evolution (DGLAP) equations down to rather small values of Q². A very likely related phenomenon is the rather large proportion of DIS events that have a large rapidity gap, a signature characteristic of diffractive scattering. This was first observed by the ZEUS experiment in the summer of 1993. Since then diffractive physics has become quite a growth industry at HERA and beyond. In photoproduction, the complexity of the photon as a particle has been confirmed, with many data now available on the different features of `direct' and `resolved' photons for both inclusive and high E_T jet physics. Rapidity gap events are also observed in the photoproduction of large E_T jets, providing important complementary information to that obtained from DIS events. Another very active topic is precisely how deep inelastic processes are related to those of real photons - data already show that the transition may take place below 2GeV₂ in photon virtuality. The performance of the HERA collider has improved steadily over the last few years. Following the first 0.03pb_-1 of data collected in 1992, around 1pb_-1 were collected in 1993, about 6.2pb_-1 in 1994 and 21.2pb_-1 in 1995. The results presented at this workshop came from the 1993 and 1994 HERA physics runs. With such a wealth of data the organizers had to be very selective in what could be covered. The main theme was the understanding (or otherwise) of QCD that the much increased phase space of HERA has given us - with particular emphasis on proton, photon and pomeron structure, the last being deliberately provocative. Many interesting results on aspects of photoproduction and high Q² physics could not be covered. The format of the workshop followed the now well established pattern of a full day of plenary talks on the Monday followed by four and a half days of intense discussion and debate in three working groups (deep inelastic scattering, photoproduction and diffraction) and ending with summaries from the three working groups on Friday afternoon. Although most of the talks at the workshop were devoted to data from HERA and its interpretation, we did have some short, but informative, talks on relevant results from the NMC, E665, CCFR and CDF experiments. The DIS group had much hard fact to contend with. Much of the discussion focused on the F₂ measurements from H1 and ZEUS using both nominal vertex data from 1993 and shifted vertex data from 1993 and 1994. The `shift' of the vertex involves a manipulation by the HERA machine physicists to cause the electron and proton bunches to intersect one bunch later in the proton direction, moving the interaction point further from the rear detectors and thus allowing a smaller electron scattering angle to be measured and hence a lower value of Q<sup>2</sup>. The surprise is that F<sub>2</sub> is still rising with decreasing x at Q<sup>2</sup> as low as 2GeV<sub>2</sub>. What has also caused intense discussion at this workshop and at conferences earlier in the summer was the demonstration that `standard' QCD could describe the data - either in the form of next-to-leading order DGLAP evolution with non-singular parton distributions at a very small starting scale in Q², as in the approach by Glück, Reya and Vogt, or in the `double asymptotic scaling' approximation to DGLAP evolution as in the approach of Ball and Forte. During the week there was much argument about what this would imply for QCD, whether indeed it could be correct, and if it wasn't why did it appear to describe the data so well. Alternative approaches based on solutions of the BFKL equation - which resums large logarithms of 1/ x rather than of Q<sup>2</sup> - are also able to describe the HERA F<sub>2</sub> data. One conclusion seems clear and that is that F<sub>2</sub> measurements by themselves will probably not be enough to settle the arguments and that some other observable such as F<sub>L</sub> or F<sub>2</sub> will have to be measured. Neither will be easy. Turning to DIS final states, the early hope that forward energy flow or forward jet rates would provide a distinctive signature for BFKL behaviour has been tempered by the reality of experimental and phenomenological difficulties of working in a region very close to the forward proton direction. Much has been learnt and experimental methods and control of systematic errors have improved greatly, but it is clear that a reliable description of forward energy flow still eludes us. At larger x and Q<sup>2</sup> good progress has been made in understanding the problems inherent in a measurement of α<sub>s</sub> from jet rates. H1 have published a measurement and the ZEUS analysis was nearing completion. A proper understanding of systematic effects in both theory and experiment is the key to what should be a well defined and reliable measurement. The topics highlighted for discussion by the plenary speakers in the photoproduction sessions were the difficulties of providing a full phenomenological description of all aspects of a photon's interactions at high energies and the advances made in both theory and experiment in high E<sub>T</sub> and jet physics. A number of problems found in the first attempts to confront theory and experiment at large E<sub>T</sub> have been overcome by the combination of much more copious data and next-to-leading order calculations. These analyses will soon be in a position to provide a useful constraint on the proton's gluon density at x values between the HERA and fixed-target F<sub>2</sub> data. Because of the propensity of a real photon to fluctuate into a hadron or a Q<sup>2</sup> pair, a large part of its interactions are similar to those of hadron - hadron interactions, thus the possibility that multiple parton interactions are required for a complete description of hard photoproduction processes must be taken seriously. The implications of this for both experiment and phenomenology were debated vigorously by the photoproduction working group. The diffractive group saw some of the liveliest discussions of the week. There are now a lot of data, both on inclusive measurements and vector meson production from both real and virtual photons. What is also exciting about this area of HERA physics is that protagonists of both hard and soft approaches believe that data (or enough of it to be interesting) support their point of view. The organizers agonised for a while about whether to allow the word `pomeron' into the title of the workshop and this summarizes the feelings of quite a number of physicists. In the end it was decided that it was a useful shorthand and also a good spark with which to light the fire of profitable argument. The inclusive data can be described in terms of a diffractive structure function. In a class of models, particularly those favoured by the proponents of the `soft' pomeron, the structure function is factorized into a pomeron flux and a function of β (the equivalent of Bjorken- x for γ-pomeron scattering) and Q². The first analyses of H1 and ZEUS data were both consistent with the soft pomeron behaviour as determined from hadron - hadron scattering. However, a later analysis by ZEUS, using a different method of non-diffractive background subtraction, shows a behaviour more reminiscent of that expected from a hard or Lipatov pomeron. There is not yet a very significant statistical disagreement. Quite a number of other phenomenologists do not see the necessity to introduce a quasi-particle at all. A week was barely enough to hear all the arguments and much activity continues amongst both theorists and experimenters. What is clear from the HERA data is a pattern of quite different behaviour of cross sections as a function of centre-of-mass energy as soon as a hard scale is present. The hard scale can be provided by Q², the mass of a vector meson, or -t, the momentum transfer squared from the proton. It is becoming clear that the soft pomeron approach will not account for all diffractive data at HERA, but we are quite a long way from a consistent picture. The feedback obtained from the participants indicated that the workshop had been a resounding success. As a result of the discussions, a number of new theoretical projects were instigated, and the exchange of ideas between theorists and experimenters will have an important influence on future data analyses. There was strong support for another similar workshop to be held in about two years time. Thanks are due to all speakers and to the organizers of the working groups: Nick Brook, Albert De Roeck, Alan Martin and Barbara Badelek (DIS); John Storrow, Aharon Levy and Chris Hilton (photoproduction); Jeff Forshaw, Julian Phillips and Halina Abramowicz (diffractive physics). Special thanks are due to the local organizer Mike Whalley and his assistant Rachel Lumpkin who coped with all emergencies with efficiency and good humour. We also acknowledge support from the UK Particle Physics and Astronomy Research Council. [1] 1993 J. Phys. G: Nucl. Part. Phys.19 1425-703