The distance to GRO J1655-40 is between 3 and 5 kpc, as determined from observations of Galactic HI toward the continuum source [Tingay et al. 1995]. The observed angular expansion of 65 
5 mas/day then corresponds to an apparent speed, relative to the speed of light, of 
. The expansion of the source is apparently superluminal.
However, what is the intrinsic speed of separation after accounting for the relativistic effects which are responsible for the appearance of apparent superluminal motion? Before answering that question some assumptions must be made about the nature of the two components seen in the images. One possibility is that one of the components is the core of the source, coincident with the source of high energy X-rays and stationary. The other component may be moving away from the core after being ejected on approximately 1994 August 13 - 14. Thus the observed expansion of the source would be due to a one-sided motion.
The second possibility is that neither of the components are the core of the source and both are in motion away from an unseen core which ejected the components simultaneously on approximately 1994 August 13 - 14. Thus the observed expansion of the source would be due to two-sided motion.
These two possibilities cannot be distinguished on the basis of an absolute registration of the four images. Absolute positions cannot be recovered from the SHEVE data for the purposes of aligning the images in celestial coordinates and no information on the component spectra are available to allow the identification of the core. Both possibilities are also consistent with an ejection event near 1994 August 13 - 14. One argument in favour of one-sided expansion is that the brightest component in each image (the south-west component) is also the most compact, reminiscent of the compact core of an extragalactic radio source.
The two possibilities cannot be distinguished on the basis of the available data. Both possibilities must, therefore, be considered. First, a one-sided expansion with the south-west component as a stationary core yields a minimum intrinsic speed for the north-east component relative to the core of between 0.7 and 0.9 times the speed of light (e.g. Pearson and Zensus 1987), for the limits on the distance to GRO J1655-40.
Second, for a two sided expansion and assuming equal component speeds and opposite directions of motion, the intrinsic component speeds are between 0.5 and 0.8 times the speed of light [Tingay et al. 1995]. If either of the possible source geometries are adopted then the intrinsic speed of ejecta forming the radio source is at least mildly relativistic.
Hjellming and Rupen [1995] have published extensive VLBA observations of GRO J1655-40 over a period of weeks, rather than days, from which they plausibly identify the core of the radio source as the south-west component which appears in the images here. They derive an intrinsic speed of approximately 0.92 times the speed of light and an angle to the line of sight of approximately 85 
. The observations presented here from 1994 August 21, 22, 23, and 24 are coincident in time with the second VLBA observation of Hjellming and Rupen [1995] and the images resulting from the SHEVE and VLBA observations agree very well.
Marked on the montage in Figure 6.5 is the position of the south-west component which has been aligned vertically. Also marked are the positions of the sub-components of the north-east component, C1, C2 and C3 and an extension to the south-west component E1. A curious aspect of the evolution in GRO J1655-40 over these four days is the rapid disappearance of the component designated C2, the middle sub-component in the north-east component. C2 is the brightest feature in the north-east component on 1994 August 21 but weakens quickly and is absent from the image on 1994 August 24 whereas the leading sub-component C1 and the trailing sub-component C3 do not fade as quickly. Also interesting is the extension to the south-west component E1, which appears to be an extension of constant length and strength from the core.
In the absence of longer term daily monitoring it is difficult to explain the rapid evolution in GRO J1655-40. Perhaps the sub-component C1 remains bright since it is the head of the jet which is advancing through the interstellar medium, similar to the head of an FR II radio source. The constant extension from the core may be the base of the relativistic jet. However, these speculations do not explain why C2 should be a highly variable feature whereas C1, C2, and E1 are not.