MOVEMENTS IN CRYSTALLIZING MAGMA ,263 



The effect of the bridging tendency in such a crystallizing magma 

 is stated in Bowen's words; the Hquid must flow "into the space 

 immediately below it, if the potential bridge is to become a reaHty." 

 It is a most significant point that the motion is at right angles to the 

 bands, each crystallizing layer contributing Hquid to form a layer 

 just below or just above. Furthermore, the deformation produced 

 in the " weak yielding segment " is from lateral compression. Both 

 the flow and the deformation would produce a vertical orientation. 

 As a matter of fact the bands in the Duluth gabbro show many 

 instances of orientation parallel to the bands and not one at right 

 angles. It should be noted also that any filter-press action at 

 Duluth must have stopped before much interlocking of crystals 

 occurred, for there is very little sign of broken or bent crystals — 

 three or four grains in hundreds of sections. There are enough 

 to show that signs of strain may be preserved, but are so few that 

 no general deformation is likely. Here, as in the previous cases 

 that Bowen outlines, the evidence of structure is emphatically 

 opposed to the action he suggests. 



Completeness of separation and volumes of the separates. — If we 

 assume, as seems probable, that Hquid may be squeezed out from 

 a crystal mesh 50-80 per cent crystalline, how complete a separation 

 can be effected ? In the banded portion of the Duluth gabbro the 

 main rocks are oHvine gabbro with only sHght changes in proportion 

 of minerals, but the bands of extreme composition are just as truly 

 bands as the rest, and are evidently formed in the same way. 

 Can an anorthosite be squeezed out of the same oHvine gabbro as a 

 peridotite ? Or must we assume that the peridotite was the residue 

 after the average gabbro Hquid was squeezed out ? If we assume 

 that peridotite is a residue, can we assume for a neighboring band 

 that anorthosite was the residue after the same average gabbro was 

 squeezed out ? 



Bowen notes that the contrast between bands should be of 

 a different order of magnitude from that shown in the gabbro- 

 granophyr association, implying that the variation in the bands is 

 relatively sHght. The bands vary from rocks 98 per cent plagioclase 

 to rocks 90 per cent magnetite, and include peridotites with no 



