A beautiful experiment of the same kind was performed by M. Thollon, of M. Bischoffsheim's observatory at Nice, in the summer of 1880.[635] He confined his attention to one delicately defined group of four lines in the orange, of which the inner pair are solar (iron) and the outer terrestrial. At the centre of the sun the intervals separating them were sensibly equal; but when the light was taken alternately from the right and left limbs, a relative shift in alternate directions of the solar, towards and from the stationary telluric rays became apparent. A parallel observation was made at Dunecht, December 14, 1883, when it was noticed that a strong iron-line in the yellow part of the solar spectrum is permanently double on the sun's eastern, but single on his western limb;[636] opposite motion-displacements bringing about this curious effect of coincidence with, and separation from, an adjacent stationary line of our own atmosphere's production, according as the spectrum is derived from the retreating or advancing margin of the solar globe. Statements of fact so precise and authoritative amount to a demonstration that results of this kind are worthy of confidence; and they already occupy an important place among astronomical data. The subtle method of which they served to assure the validity was employed in 1887-9 by M. Dunér to test and extend Carrington's and Spörer's conclusions as to the anomalous nature of the sun's axial movement.[637] His observations for the purpose, made with a fine diffraction-spectroscope, just then mounted at the observatory of Upsala, were published in 1891.[638] Their upshot was to confirm and widen the law of retardation with increasing latitude derived from the progressive motions of spots. Determinations made within 15° of the pole, consequently far beyond the region of spots, gave a rotation-period of 38-1/2, that of the equatorial belt being of 25-1/2 days. Spots near the equator indeed complete their rounds in a period shorter by at least half a day; and proportionate differences were found to exist elsewhere in corresponding latitudes; but Dunér's observations, it must be remembered, apply to a distinct part of the complex solar machine from the disturbed photospheric surface. It is amply possible that the absorptive strata producing the Fraunhofer lines, significant, by their varying displacements at either limb, of the inferred varying rates of rotation, may gyrate more slowly than the spot-generating level. Moreover, faculæ appear to move at a quicker pace than either;[639] so that we have, for three solar formations, three different periods of average rotation, the shortest of which belongs to the faculæ, one of intermediate length to the spots, and the most protracted to the reversing layer. All, however, agree in lengthening progressively from the equator towards the poles. Professor Holden aptly compared the sun to "a vast whirlpool where the velocities of rotation