The displacement may be due to two causes :sx - ( 1 ) Shrinkage of the refractories , which takes place apparently on account of the action of metallic vapour , even though the refractories have been previously fired at temperatures above those reached in the experiments ( this is referred to later) .sx ( 2 ) After immersion in the molten metal the tube tends to lift on account of its buoyancy ; this causes the axis of the tube to deviate from the vertical and results in difficulty in alignment of the pyrometer in subsequent observations , while serious bending stresses are set up which shift the relative position of the " matching " area still further .sx These observations indicate not only that the refractories must be of high purity , which has the disadvantage that they are apt to be very friable , but that they must also be as hard as possible so that the pots and tubes shall be strong and will not distort in use .sx The importance of hard refractories for this work cannot be over emphasized .sx The difficulty of shrinkage was overcome by the use of pure oxides suitably ground and fired at high temperatures .sx The difficulty of maintaining the tubes in place was overcome in two ways ( 1 ) by using a moulded " head " or block formed of a rigid mass of refractory in which the tube was held , and ( 2 ) by using tubes moulded as part of the pot , projecting upwards from the base , thus avoiding any discontinuity in the base of the crucible .sx The latter arrangement gives the tube more rigid support .sx Pure oxides such as alumina , thoria , magnesia , etc. , are not readily formed into crucibles , even of the simplest shape .sx The design of black-body enclosures is further largely influenced by the necessity of providing for the removal of the very tender freshly-pressed pot from the mould in which it has been formed .sx In the present design of refractories ( see Figs .sx 2 , 3 and 4 ) the diameter of the immersed tube is large in comparison with the diameter of the orifice in the baffle plugs , which in no case exceeds one-eighth of the depth of immersion .sx The effect of the depth of immersion of the refractory tube in the optical method is rather different from that in the thermocouple method .sx In the latter case the temperature of the metal adjacent to the tip of the thermocouple wires is being measured , and it is only necessary to prevent conduction along the wires apart from the surface loss by radiation .sx In the case of the optical method , however , the observed temperature is the average temperature of a relatively large enclosure ; and when once black-body conditions have been attained by a sufficient depth of immersion , further immersion does not improve the conditions of measurement .sx The insertion of a baffle plug facilitates the attainment of black-body conditions without employing large depths of immersion .sx For example , the reduction of the apparent temperature by departure from black-body conditions in the absence of a baffle plug was illustrated in the case of an experiment upon iron .sx The mean temperatures of melting and freezing were found to be as low as 1,507 C. and 1,504 C. respectively , when observations were made sighting either into the immersed head as shown in Fig. 2 , or from above into the top of the tube held in the .sx centre of the large crucible as shown in Fig. 3 .sx When baffle plugs were inserted , the observed mean temperatures were raised to 1,527 C. .sx Considerable attention has been devoted over a period of some years to the development in the Metallurgy Department of pure refractories in connection with the metallurgical researches in progress .sx This work has been in the charge of MR. D. TURNER , to whom the authors are indebted for the preparation of the ground material and the firing of the refractories , as well as for co-operation in the production of the pots .sx The pure alumina used for the preparation of the refractories in the present research was obtained from the British Aluminium Company , Ltd. The analysis of this material showed that the only impurity present was .sx a trace of iron , less than 0.01%. The material , which in the condition as received passed through a 120-mesh sieve , was ground to a very fine powder in an end runner mill .sx The powder was then magnetically separated to remove any traces of iron which might have been introduced by grinding , and entirely passed through a 200-mesh sieve .sx After being made damp .sx with water it was pressed into the mould by gradual filling and repeated ramming .sx The chief difficulty lay in the successful extraction of the pot from the mould .sx It was essential that all the portions of the refractory object were sustained in slight compression during the extraction process .sx The mould for making the refractory head marked B in Fig. 2 opens in three pieces to allow the head to be removed .sx The central tube is formed by a metal pin which is withdrawn before the mould is opened .sx In the case of the pot shown in Fig. 3 , in which the central tube is rigidly fixed , the shape of the internal ring is formed by a polished brass core around which the material is packed .sx The arrangement of the mould and core is shown in Fig. 5 .sx Great care is needed in extracting this core ; for this purpose the whole mould is held in a screw press which exerts a downward push on the top of the central tube while the core is being pulled out .sx The pot is then removed from the main cylinder by pushing .sx the base plate upwards .sx The centre pin 3 is then pulled out before lifting the pot , from the base .sx The crucibles are then fired at a temperature of 1,530 C. to 1,560 C. .sx The types of crucibles , tubes and heads and their respective " baffles " are shown in section in Figs .sx 2 to 4 inclusive .sx In the form shown in Fig. 2 the immersed tube C constitutes a part of the block of refractory material B which rests in the crucible .sx The mass of this block is sufficient to keep the tube in position .sx The molten metal E is contained in a hard refractory crucible F whose capacity is normally about 8 ml .sx The tube C which is held centrally in the pot contains a perforated refractory plug D which acts as a " baffle " and prevents excessive loss of heat by radiation .sx This baffle plug has a small axial aperture which is reduced in diameter by steps so as to facilitate the sighting of the pyrometer .sx Considerable difficulty was experienced in the early work on account of the fracturing of the tip C at the sharp junction which it formed with the main block B. This was remedied later by the provision of a suitable curvature at this change of section .sx In the arrangement already described , and also in that shown in Fig. 3 , the optical pyrometer is sighted from above .sx In the design shown in Fig. 3 the black-body chamber is formed by the insertion of the baffle plug in the central tube which forms a part of the crucible .sx This tube passes through the entire depth of the pot , but before use the lower half is tightly packed with powdered alumina from the bottom to within 1 in .sx from the top of the crucible .sx The capacity of this crucible is about 24 ml in normal working conditions .sx In the form of apparatus shown in Fig. 4 the pyrometer is sighted from below the crucible .sx The tube whose closed end is totally immersed in the molten metal projects upwards about in .sx from the bottom of the crucible .sx The " baffle " used in this pot is the same as that shown in Fig. 3 and is held in position by a refractory block containing a slightly smaller hole than that in the crucible .sx The same form of crucible tube and baffle was also used in a somewhat smaller size .sx In this research two effects due to the presence of metallic vapour have been observed ( 1 ) the shrinkage of refractories , and ( 2 ) the absorption of light by metallic vapour , resulting in the lowering of the observed melting points .sx These effects may have been only slight in previous investigations , possibly owing to the fact that the refractory black-body chambers used in the course of those determinations were of such a compact nature that the diffusion of metallic vapour through their walls was inhibited .sx In the present research , however , it was necessary to use refractory bodies of a special nature , and these were somewhat porous ; consequently metallic vapours could diffuse through them more readily .sx This was clearly shown , for instance , when a black-body head was examined after it had , been immersed in gold and kept just below its melting point for several hours .sx The colour of both the tube of the head and the baffle plugs inside was changed from white to a pinkish tinge , which suggested the possibility of the condensation of fine metallic particles from previously existing vapour .sx Furthermore , .sx baffle plugs made from refractories containing a certain amount of siliceous material were found to shrink during use in the presence of iron vapour , while baffle plugs made of pure hard alumina showed no visible shrinkage in similar circumstances .sx This suggests that the metallic vapour is a possible cause of shrinkage .sx The second effect was first noticed during the course of some observations on the melting point of gold .sx In this determination the small-sized black-body head ( see Fig. 2 ) was used and six observations were made , three on cooling and three on heating through the melting point .sx The individual temperatures of the arrests which were taken on cooling and heating respectively agreed amongst each other to within less than 0.5 C. , whereas the mean value of the results observed on cooling through the melting point was 1.5 C. lower than that on heating .sx The apparatus in which this experiment was carried out was then kept for a further 24 hours at a temperature just below the melting point of gold , when further observations were made .sx During six observations , each after an interval of 24 hours , the melting point became successively lowered .sx The difference , for example , between the temperature of the first and last values obtained by cooling through the melting point was 4 C. approximately , and by heating through the melting point 1.5 C. .sx In order to check the above results , further experiments were carried out using a pot of the type shown in Fig. 3 .sx The gold was kept molten for a period of two days except for the purpose of observation during freezing and melting .sx The temperature at which fusion occurred was observed at intervals of 24 hours and 48 hours after the initial melting and freezing .sx The results are plotted in Fig. 10 .sx The temperature of the melting of the gold was found to be depressed 1.8 C. when observations were taken 24 hours after the initial readings , then a further 4 C. after 48 hours , while the temperature of freezing was depressed 1.8 C. and a further 2 C. in similar periods .sx The actual presence of vapour in the black-body chamber does not affect the apparent temperature , but if the vapour passes into the sighting tube above the black-body chamber , which is at a slightly lower temperature , then condensation may occur .sx The fog caused by this condensation consists of small metallic particles at a different temperature from that of the black-body , and consequently the apparent temperature will be in error .sx Hence in the above case the vapour from the gold must have formed a fog in the slightly cooler portion of the " sighting " tube .sx On examination of the black-body head after these two sets of experiments it was noticed that instead of being white , the colour of the refractory had become a pale pink , and that this coloration penetrated through the material so that the baffles were also coloured .sx This confirmed the impression that gold particles were the cause of the apparent lowering of the melting point of gold , for the black-body conditions were otherwise excellent .sx