Rats are versatile animals, and are very adept at escape from confinement, in particular, when under the influence of stimulant drug. Lost data due to animals escaping from the apparatus were a significant problem when using the open hemispheric bowls used in early designs. One approach is to try and fit a lid, or to design a globe-shaped bowl with only a small opening for the wire at the top of the sphere. In each case, ease of putting the animals in the bowls and removing them afterwards (especially, if retrieving a still hyperactive animal) becomes an issue. Third is to use a cylinder with high straight sides, which combines difficulty of escape and ease of changing animals. This can result in excessive flexing of the tether wire between times when the animal rears and when it extends into the outer corners of the floor, with an increased chance of the animal getting tangled in the tether. In our laboratory, we use polystyrene mixing bowls, containing a handful of sawdust, fitting snugly inside a tall transparent Perspex cylinder (Fig. 1b). This allows a good compromise for maintaining a centre of curvature, ease of change and ease of cleaning and changing sawdust (including the use of different materials for environmental conditioning and sensitisation of rotational responses). The reason for using a transparent cylinder is so that the animals’ behaviour can be monitored by observation or video, in particular, to allow rating of l-dopa- and graft-induced dyskinesias simultaneous with recording rotation behaviour, under alternative pharmacological and cell-based treatment regimes (14) (see also Lindgren and Lane, this volume).

The harness must be both comfortable for the animal, not restricting its movements, but also designed to avoid escape. Early systems involve a bent or curved wire belt, positioned tightly around the chest behind the elbows. However, these are difficult to fix reliably so that the animal cannot escape without over-tightening. The most secure systems are those involving a fabric jacket with holes to accommodate the forelimbs, and generally closed using Velcro. These are difficult for the animal to escape from, but can be quite fiddly or time consuming to fit, especially, if needing to change multiple animals in a batch between runs. A compromise between these two approaches is to use a broad fabric belt passed around the chest behind the elbows and fixed with velcro fastening for a firm, but not overtight adjustment. The harness or jacket may either be permanently fixed to the tether wire, in which case the animal must be placed in the harness while standing within reach of the tether wire very close to the rotometer, or the harness may be attached by a removable connector (such as a syringe type Luer connector) which then allows the rats to be harnessed more conveniently at a nearby bench top. In our laboratory, we have adopted a slightly different strategy, using a large rubber band (US: elastic band) – in the UK the standard Post Office band for bundling letters is ideal (Fig. 1b). The band is simply and efficiently applied by passing around the tips of the five fingers of one hand, fanning opening the hand to stretch the band, quickly passing over the head of the rat held in the other hand, and positioning snugly behind the elbows. The band is then connected to the tether by a small crocodile clip soldered to the end of the tether wire to allow a low pressure-elasticated grip, which can be adjusted to the size of the rat by pinching a fold of the band in the clip.

Free turning of the pivot assembly is essential, and this joint should be a ball race to allow long-term free turning without resistance. Equally, the tether wire has to be reasonably flexible, but with no opportunity for twisting; piano wire is frequently used. Early assemblies used a cam and mechanical switch assembly, but this is better replaced by an optical switch that again applies no resistance. The most simple systems involve a single switch to detect 360° turns, but modern systems invariably involve the ability to detect smaller angles of turn, for example, by recording photo-beam interruptions by multiple holes or notches in a turning disc. There is a distinct advantage in a system with two closely adjacent switches or sensors that allows the direction of turning to be distinguished and recorded separately. In our laboratory, although our system has the capacity to record quarter turns, drug-induced rotation in 6-OHDA-lesioned animals is reliable, consistent and almost exclusively in one direction, so we routinely collect only data on 360° turns, separately for ipsilateral and contralateral direction, and then analyse the data in terms of net difference measures. However, there are occasions where it may be relevant to record quarter turns or to analyse data separately for the two directions, in particular, when looking at the effects of spontaneous turning, small lesions, intact animals, stereotyped responses, or other lesions with as yet uncharacterised turning morphology.

Early systems used separate electromechanical counters attached to each pivot assembly, with manual recording of data registers. Nowadays, rotometer control systems are almost exclusively microelectronic, typically with a small laptop computer or dedicated ­processor receiving inputs from an interface panel connected to multiple rotometer pivot heads. The choice of hardware and ­software is endless; in established behavioural laboratories, the ­logical solution is to extend other standard control equipment to cover rotometer data recording; alternatively, the alternative commercial systems will provide the control unit and associated software, and this typically constitutes a significant proportion of the total system cost. Rotation experiments often involve some dozens of experimental animals, often with the individual drug tests running for several hours, and experimental designs often involving multiple tests on separate days. Consequently, it is considerably more efficient to test and run animals in a bank of rotometer bowls, rather than individually. So how many systems should be ­purchased/constructed and run in parallel? As a rule of thumb, a bank of eight is suitable for a small laboratory running rotation experiments on an occasional or regular basis. Just purchasing four bowls would do, but then most of the cost is taken up in the control equipment. In our laboratory, we run a bank of 16 rotometers from one ­control computer, since we frequently have ten or more people engaged in long-term experiments with rotation as a primary readout. A ­significant factor in scaling up is the time it takes to change ­animals between test runs. With a relatively slow onset and long-acting drug such as amphetamine, it is no problem setting up the batch software, changing eight animals and then hitting the run key. However, with 16 changes which may take up to 5–10 min, or with a rapid onset and relatively short acting drug (such as the mixed dopamine receptor agonist apomorphine), then it is better to have hardware and software designed to start each bowl separately (perhaps with a remote trigger) within the same batch run. The capacity of some systems to provide for 32 or more parallel inputs is really only applicable to a pharmaceutical scale of high-throughput screening.

With the rise in a range of genetic models of disease, in particular in mice, there is considerable emerging interest in methods to evaluate motor behaviours, including rotation, in mice (15) (see also Brooks or Smith and Heuer, this volume). Following unilateral 6-OHDA lesions, mice rotate in response to the same stimulant drugs as do rats (16, 17). Indeed, most equipment suppliers offer scaled down bowls, jackets and tethers for mice. However, mice are not just little rats. It has proved more difficult to achieve stable and reliable unilateral lesions in mice, it is considerably more difficult physically to fit them with rotation harnesses, and they are much more efficient at escaping, getting tangled or reacting badly to the constraint. So just because multiple mouse rotometers are available on the market, it must not be assumed that they work reliably or well. Our advice would be to only purchase such a device if you have clear-cut independent evidence from an experienced mouse behaviour laboratory that the particular system does work without complication as advertised. In our laboratory, we have sought to adapt two consecutive systems to mice without great success. Consequently, for our current experiments in mice, we have reverted to an updated variant on the manual observation and sampling strategy (see Sect. 2.1), by video recording the test sessions with the treated mice placed without harnesses in our existing bowls and transparent cylinders, and then manually documenting individual animals’ turning on ten times fast forward video play back (see Heuer and Smith, this volume).

Although most rotation monitors have been based on the rotometer principle of mechanical recording of turning in tethered animals, a number of alternative systems developed for locomotor tracking and movement analysis have employed software to detect and quantify the turning behaviour and rotation of freely moving animals in open arenas. The techniques of animal tracking can involve a number of different technologies. Ethovision XT and the Panlab SMART systems (see Table 1) involve video analysis of locomotor paths using image analysis to detect the location of the rat by contrast against its background, to determine the point centre of gravity, and to record locomotor tracks within the test arena. The location of the rat in the arena can, alternatively, be monitored by movement of the animal through an electromagnetic field or by breaks in a Cartesian array of photocell detectors (Med Associate, Table 1). Although such open field systems are largely used to characterise parameters such as locomotor activity, speed of movement, stereotyped run paths, rearing, etc., “rotation” can usually be extracted by software algorithms analysing locomotor paths in small or wide angled circles. In most such systems explicit turning of the body axis rather than turning of the point centre of gravity in space is difficult, although the Ethovision TriWise software does permit inference of true rotation by extracting a three point head-back-rump longitudinal axis from the recorded images. Such systems have much broader application to a wide range of behavioural classes, but they are not as reliabile, precise or as easy to use as a mechanical rotometer, when the specific purpose of the research programme is to use the rotational asymmetry of unilateral lesioned animals as the primary outcome in designed experiments.