The potential of bulk melt-processed YBCO single domains to trap significant magneticfields (Tomita and Murakami 2003 Nature 421 517–20; Fuchs et al 2000 Appl. Phys. Lett. 762107–9) at cryogenic temperatures makes them particularly attractive for a variety ofengineering applications including superconducting magnets, magnetic bearings andmotors (Coombs et al 1999 IEEE Trans. Appl. Supercond. 9 968–71; Coombs et al 2005IEEE Trans. Appl. Supercond. 15 2312–5). It has already been shown that large fields can beobtained in single domain samples at 77 K. A range of possible applications exist in thedesign of high power density electric motors (Jiang et al 2006 Supercond. Sci. Technol.19 1164–8). Before such devices can be created a major problem needs to beovercome. Even though all of these devices use a superconductor in the role of apermanent magnet and even though the superconductor can trap potentially hugemagnetic fields (greater than 10 T) the problem is how to induce the magneticfields. There are four possible known methods: (1) cooling in field; (2) zero fieldcooling, followed by slowly applied field; (3) pulse magnetization; (4) flux pumping.Any of these methods could be used to magnetize the superconductor and this may bedone either in situ or ex situ.Ideally the superconductors are magnetized in situ. There are several reasons for this: first,if the superconductors should become demagnetized through (i) flux creep, (ii)repeatedly applied perpendicular fields (Vanderbemden et al 2007 Phys. Rev. B 75(17)) or (iii) by loss of cooling then they may be re-magnetized without the needto disassemble the machine; secondly, there are difficulties with handling verystrongly magnetized material at cryogenic temperatures when assembling themachine; thirdly, ex situ methods would require the machine to be assembledboth cold and pre-magnetized and would offer significant design difficulties. Untilroom temperature superconductors can be prepared, the most efficient design ofmachine will therefore be one in which an in situ magnetizing fixture is included.The first three methods all require a solenoid which can be switched on and off. In the firstmethod an applied magnetic field is required equal to the required magnetic field,whilst the second and third approaches require fields at least two times greater.The final method, however, offers significant advantages since it achieves thefinal required field by repeated applications of a small field and can utilize apermanent magnet (Coombs 2007 British Patent GB2431519 granted 2007-09-26).If we wish to pulse a field using, say, a 10 T magnet to magnetize a30 mm × 10 mm sample then we can work out how big the solenoid needs to be. If it werepossible to wind an appropriate coil using YBCO tape then, assuming anIc of 70 A and athickness of 100 µm, we would have 100 turns and 7000 A turns. This would produce aB field ofapproximately 7000/(20 × 10−3) × 4π × 10−7 = 0.4 T. To produce 10 T would require pulsing to 1400 A! An alternative calculation would be to assume aJc of say5 × 108A m−1 and a coil1 cm2 in cross section. Thefield would then be 5 × 108 × 10−2 × (2 × 4π × 10−7) = 10 T. Clearly if the magnetization fixture is not to occupy more room than the puck itselfthen a very high activation current would be required and either constraint makes in situmagnetization a very difficult proposition.What is required for in situ magnetization is a magnetization method in whicha relatively small field of the order of millitesla repeatedly applied is used tomagnetize the superconductor. This paper describes a novel method for achieving this.