Search Documents

By S. Lubbe

South Africa and Australia are the biggest suppliers of zircon sand to the international zirconium industry. However neither South Africa nor Australia is well known for zircon beneficiation. Geratech Zirconium Beneficiation Ltd (GZB) continued with additional research on sodium hydroxide (NaOH) cracking of zircon sand during 2002-2003. In 2003 GZB started extracting zirconium from zircon sand by means of NaOH cracking on a commercial scale. Experience has shown that temperature profile and atmospheric control inside the furnace is crucial for the beneficiation of zircon sand. Silica carryover to zirconium chemicals could result if a high temperature is used. Once the sodium silicate is extracted from the sodium zirconate and dissolved in hydrochloric acid, two distinct routes can be followed to precipitate various zirconium chemicals. The most common route is to precipitate zirconium oxychloride crystals (ZOC), with subsequent purification from all contaminants (crystal route). Less known is the process (liquid route) that involves the direct precipitation of zirconium basic sulphate (ZBS). This route will yield a less pure product, with contaminants such as silica and titanium. An important factor in this route is the prevention of silica gel formation, which could hamper final product filtration. For applications like paint drying (zirconium octoate) or antiperspirants (zirconium hydroxychloride) low levels of contaminants have no effect on the final product. The advantage of the liquid route is fewer production stages compared to the crystal route. The disadvantage of the liquid route is that the market for the products will be significantly smaller. The optimum solution could be a plant design that could cater for both routes. Another example of an application of zirconium chemicals is the use of ammonium zirconium carbonate (AZC) in the paper industry. Zirconium basic carbonate (ZBC) is dissolved in ammonium carbonate to produce AZC solution. AZC is used mainly in European countries in the paper industry. For example, carton boxes were initially produced with formaldehyde as the binder, however, it has now been replaced with AZC since formaldehyde is considered toxic. AZC reacts with the cellulose fibres in the paper to act as the binder. The resulting product is not toxic, and printing ink dries very quickly due to the porous paper structure. Other applications of zirconium chemicals involve the use of acid zirconium sulphate tetrahydrate (AZST), zirconium orthosulphate (ZOS), potassium zirconium carbonate (KZC), and zirconium hydrous oxide (ZHO). Fluoride-based zirconium chemicals like zirconium tetra-fluoride (ZrF4) and hexafluoro-zirconic acid (H2ZrF6) are used in the aluminium industry. Zirconium oxide (Zr02) can be produced from any of the abovementioned precipitated chemicals via a high-temperature decomposition process. The physical properties of such oxides can differ tremendously, depending on the final application. The fired density of sanitaryware is typically 5.8 g/cm3, compared to milling media at >6.1 g/cm3. The required fired density is achieved by controlling the precipitation and decomposition conditions for these two oxides. The same applies to stabilized and mixed oxides, for example where zirconium oxide acts as an oxygen carrier in fuel cells.

Jan 1, 2012

By AIME AIME

[SECRETARY'S NOTE.-Since the Committee on Publication of United Engineering Society, will issue a memorial volume re¬porting these proceedings in full, an outline only will be here given. The sessions were held in the Main Auditorium.¬R. W. R.] COMMITTEES. EXECUTIVE.-J. W. Lieb, Jr., Chairman; C. W. Rice, Secretary; John Hays Hammond, Chas. Wallace Hunt, F. R. Hutton, Charles Kirchhoff, T. C. Martin, E. E. Olcott, Chas. F. Scott, Samuel Sheldon, Ambrose Swasey. INVITATIONS.-R. W. Raymond, Chairman; Charles Kirchhoff, Secretary; A; C. Humphreys, Robert W. Hunt, L. B. Stillwell. PROGRAM.-A. R. Ledoux, Chairman; C. W. Rice, Secretary; Theodore Dwight, Chas. Wallace Hunt, T. C. Martin, Ralph W. Pope, Joseph Struthers. WAYS AND MEANS.-James Douglas, Chairman A. R. Ledoux, Secretary; H. R. Brown, J.-M. Dodge, C. L. Edgar, John Hays Hammond, W. M. McFarland, E. E. Olcott, Samuel Sheldon, J. G. White. HISTORICAL AND PUBLICATION PRESS.-T. C. Martin, Chairman; Albert Spies, H. H. Suplee.

May 1, 1907

By Deno M. Pappas

This report presents results of a U.S. Bureau of Mines study of longwall gob material. The objective of this work was to determine material stiffness properties of the gob for use in numerical models of rock mass response to longwall mining, Photographs of actual mine gob were digitized to obtain approximate particle size gradations of gob material. The gradation curve was shifted down to a laboratory scale, and 20 uniaxial compression tests were conducted. Varying the maximum particle size was not found to affect the stress-strain behavior, but changing the gradation appeared to influence the stress-strain behavior. The stress-strain relationship of the simulated gob material was nonlinear, the stress-secant-modulus relationship was approximately linear, and the stress-tangent-modulus relationship was approximately a second-order polynomial function. Equations were generated from these curves, providing numerical modelers with a means to estimate gob moduli based on the stress level. In addition, the experimental data were statistically evaluated using multiple regression analyses, producing a series of equations to predict the secant and tangent moduli from the given stress level, bulking factor, rock strength, and thickness-to-width shape ratio of the particles.

Jan 1, 2010

There is wide agreement that sustainability has multiple dimensions, but substantially less agreement on the appropriate balance among those dimensions. This occurs because peoples’ vision of a sustainable future, their goals for sustainable development, and their objectives for resource management are reflections of their values, culture, and life experiences. Even assuming agreement about the objectives of sustainable resource management can be reached; there is still a need for policy and decision makers to choose from among alternative courses of action, each of which has its own set of impacts and outcomes. One of several tools used to compare the consequences of alternative actions is the Computable General Equilibrium (CGE) model. This paper uses CGE modeling to demonstrate how policy decisions do not take place in a vacuum and how policy-makers must be aware that there are trade-offs involved in any decision.

Jan 1, 2005

By Mohamad Hussein

Low strain dynamic tests are performed to evaluate the structural integrity of driven or cast-in-place concrete piles. This type of testing requires minimal pile preparation, and is simply performed by attaching an accelerometer to the pile head and striking the pile head with a hand held hammer. The measured data is processed by a dedicated system, inspected for wave reflections, and interpreted based on one dimensional wave propagation theory. Piles are normally tested shortly after installation so that problems can be identified early and corrective measures taken before construction of the superstructure. However, evaluation of existing foundations is occasionally required. Often, pile installation records are not available rendering impossible any rational foundation reanalysis. This paper discusses three case histories where low strain dynamic measurements allowed determination of pile lengths and structural integrity under existing structures where pile heads were not accessible.

Jan 1, 1992

Jan 1, 1970

By Detnet Solutions

1. “Daisy-Chain” concept The term “daisy-chain” stems from the fact that when the HotShot? detonators are connected there is a chain or electrical path for electrical current to flow from detonator A to detonator B, detonator B to detonator C etc until the path is terminated by an electrical means. The “daisy-chain” concept has long been used in electronic detonator systems, historically used in Expert Explosives’ ExEx 1000 5/6 wire system and currently in use in African Explosives Limited (AEL’s) Electrodet? electronic detonator system and has been adopted as the basis upon which the “auto-programmability” of the HotShot? electronic detonator system functions. HotShot? Detonators are connected to each other in a string, completing a “daisy chain”, and during the auto-detection mode of the control equipment, it is possible for the equipment to determine the order of hookup of the detonators as well as map the entire blast layout. The auxiliary components (Row Controllers & Branch Controllers) assist the control equipment in mapping the entire blast by indicating different points on the blast, i.e. The inflection point of a row of blast holes on a closed chevron timing pattern, the beginning of row of blast holes on open chevron timing pattern and extra blast holes or extra blast rows.

Jan 1, 2004

By Atsushi Shibayama, Toyohisa Fujita, Wan-Tai Yen, Ke-jun Liu

"A gold has been leached with ammonium thiosulfate solution and several leaching conditions have been studied. The leached solution has been recovered by solvent extraction and the optimum conditions have been investigated.SUMMARYIn this investigation, a sponge gold sample (99.92% pure) has been leached with ammonium thiosulfate solution. The effects of CuSO4, (NH4)2S2O3, (NH4)2SO4, NH4OH and retention time on the gold leaching % (extraction %) and thiosulfate oxidation % (consumption %) have been studied. The copper ion acts as a catalyst in gold dissolution. On the other hand, it also accelerates oxidation of thiosulfate (consumption). It was found in this study that the best gold extraction condition was obtained at 43.2 ks with 0.8 – 1.0x103 mol/m3 (NH4)2S2O3, 30 mol/m3 CuSO4, 2.0 – 3.0x103 mol/m3 NH4OH, 0.4 – 0.5x103 mol/m3 (NH4)2SO4, pH 10±0.2 and 150 rpm. The thiosulfate oxidation was about 15% of the initial concentration."

Jan 1, 2003

Peculiarities of Quartz Distribution in Zinc Flotation Products and their Influence to Mineral Processing

Sep 13, 2010

By Keith E. Meyers

Condition monitoring is a viable troubleshooting practice among many industries to extend equipment performance, minimize unplanned downtime, reduce maintenance and repair time and costs, and increase related aspects of plant safety. But obstacles abound for mining, mineral-processing and cement operations striving to implement the technologies. Harsh machine environments can muddy efforts. High vibration in machinery can interfere with wide-spread rollout. Staff expertise or skills may be lacking. And costs may be prohibitive. Fault detection has emerged as a practical and cost-effective solution to identify process and machinery faults before failure interrupts productivity or jeopardizes worker safety. Fault detection can be a stand-alone system or connected to plant automation.

Jan 1, 2005

By A. Mezei

A process has been developed to recover cobalt, gold and bismuth from the NICO deposit in Canada. The process consists of a bulk rougher flotation followed by an optional separation by flotation to produce a bismuth concentrate and a cobalt concentrate, grading on average 12.8 % Bi and 1.8% Co, respectively. Cobalt is leached by pressure oxidation at 97% efficiency, producing a pregnant solution suitable for further processing. Before optimization, about 38% of the bismuth is recovered in the selective concentrate, wherefrom it is recovered by ferric chloride leaching followed by cementation, at overall efficiencies exceeding 98%. The gold is recovered from the cobalt and bismuth leach residues by cyanidation, at 95% efficiency. Previous work described the results of the leaching of the concentrates, recovery of gold (Part I) and the recovery of bismuth (Part 11). This work describes the results of the cobalt recovery testwork and the resulting conceptual flowsheet. The sulfur dioxide-oxygen system was the most efficient for the oxidationlprecipitation of iron and arsenic from cobalt bearing pregnant leach solutions, producing purified solutions assaying <1 mg/L As and <8 mg/L Fe. The efficiency of cobalt solvent extraction (from the resulting purified PLS containing about 1 g/L Co) ranged from 9 1 to 98%; the co-extractions ranged from 88 to 100% for copper, from 28 to 33% for magnesium, from 6 to 7% for calcium and from 96 to 98% for zinc. Sulphide precipitation processes (using either sulfur dioxidelelemental sulfur or sodium sulfide) produced residual copper concentrations in the cobalt strip solution ranging from 0.1 to 2 mg/L Cu. About 90% of the calcium and magnesium were separated by scrubbing the organic with cobalt sulfate, producing residual concentrations of 12 mg/L Ca and from 180 to 330 mg/L Mg in typical feed solutions directed to cobalt carbonate precipitation tests. Cobalt carbonate precipitation efficiencies ranged from 75 to 78% when calcium and magnesium co- precipitations were restricted from 8 to 17% and 0.3 to 0.4%, respectively. The limits required by most stringent specifications on commercial grade cobalt carbonate products were met or exceeded.

Jan 1, 2002

By P. Niessen, W. C. Winegard, C. J. Beingessner

DURING the last 10 years, several grain-growth studies have been made using zone-refined metals with controlled amounts of solute to elucidate the interaction of solute with moving boundaries. In all these studies, it was more or less implicitly assumed that the alloys used were, after recrystallization and grain growth, homogeneous within the grains with respect to solute and lattice defects. This note deals with observations made on microstructures originating from solidification of dilute alloys and from grain growth. Briefly, the experimental procedure used in these studies was the following: zone-refined cadmium and master alloys were melted and cast in Pyrex under vacuum to produce bars 14 mm in diam and about 150 mm long. These bars were cut into 8-mm-thick slices which were mechanically and chemically polished. The specimens were then deformed 30 pct in compression and given a 1-week anneal at 260°C in argon. After this anneal, a second deformation produced completely recrystallized samples 4 mm thick. The time-temperature controlled anneals were carried out in a bath of silicone oil. After the grain-growth anneal, the specimens were polished in Gil-man&apos;s&apos; solution (33 g CrOs, 6 g NazSO, in 100 ml of water). Several etches were used depending on the alloy investigated, but mostly a mixture of concentrated HC1 and HNOj (98:2 vol) was employed. In all the alloys examined, the ease with which substructures could be revealed depended strongly on the crystallographic orientation of the surfaces of the grains. However, very frequently the microstructures to be discussed here etched up over broad areas comprising several grains. Fig. 1 is a micrograph of a Cd-0.05 at. pct Mg alloy annealed for 100 sec at 270°C. Different substructures are seen in this photograph. A denotes the now existing grain boundaries. The well-defined lines and points marked B are the remains of a solidification substructure. The lines are boundaries between cell colonies, and the points are nodes located on regular two-dimensional cells. This cellular structure has the same appearance as that found by weinberg2 for rapidly quenched Zn-Au alloys. The "ghost boundaries", marked C, are very similar in appearance to a grain boundary array in a polycrystalline sample. Repeated polishing and etching showed that all the substructures observed were three-dimensional. To date, the ghost boundaries have been clearly observed in all samples containing 0.005, 0.01, 0.05, and 1 at. pct Mg, and in samples containing 0.01 at. pct Zn. Attempts to outline these ghost boundaries in cadmium alloys containing 0.01 at. pct Ag and Au were only partially successful. In these alloys, parallel lines, having a resemblance to the ghost boundaries in the Cd-Mg and Cd-Zn alloys, were observed, but it was not possible to obtain complete arrays of ghost boundaries which extended over several grains. It is possible that a different etch than that presently employed would reveal the ghost boundaries in the Cd-Au and Cd-Ag alloys as clearly as in the Cd-Zn and Cd-Mg alloys. A close examination of the ghost structure showed that it consisted of linear etch-pit arrays, Fig. 2. The ghost boundaries were resolvable even in grains which showed over-all faceting due to the etching, as seen in the corner of Fig. 2. The general appearance of the ghost structure suggests that it originated during grain growth. This is supported by the observation that the mesh width in the ghost-boundary array generally increased with increasing final grain size. To test whether the ghost structure was formed during the long homogenizing anneal or during the grain-growth anneal, some as-cast samples were deformed and given a grain-growth anneal. This anneal was terminated by air cooling or

Jan 1, 1967

By Luo Yang

A one-dimensional wave equation based algorithm for estimation of shaft and toe resistance of driven piles using High Strain Testing (HST) data is presented. The pile is represented by a uniform elastic bar, while the soil around the pile is assumed as a homogenized isotropic medium, and Smith model is adopted to represent soil-pile interaction in the wave equation analysis. In most cases, the displacement of pile particles along pile shaft is greater than recoverable soil deformation (i.e., soil quake in Smith model). Therefore, static shaft resistance could be reasonably assumed to be totally mobilized during pile driving. Smith damping factor is used to account for combined dynamic and rate effect. Force and velocity time histories measured at the pile head are used as input boundary conditions in an analytical solution of the one-dimensional wave equation. As a result, two unknown parameters, static shaft resistance and Smith damping factor, can be incorporated into the one-dimensional wave equation and determined analytically for the time duration prior to the first downward wave reaching the pile toe. Then, soil resistance and Smith damping factor at pile toe can be further determined. The advantage of the proposed analysis technique is that the Smith damping factor and the static soil resistance can be directly determined based on the closed-form solution of the one-dimensional wave equation, which is computationally efficient. Numerical examples are given to illustrate the use of the proposed method to determine the pertinent Smith model constants and the static resistances.

Jan 1, 2006

University of Alabama University, Alabama Mining and Metallurgical Society WILLIAM R. HIGGS, President ROBERT E. MEAD, Secretary J. R. CUDWORTH, Faculty Sponsor MILTON H. FIBS, Counselor University of Arizona Tucson, Arizona Miners&apos; Society LEE BLESSING, President R. W. CORY, Secretary E. P. MATHEWSON, Faculty Sponsor B. R. HATCHER, Counselor University of California at Los Angeles Los Angeles, California Student Chapter, A.I.M.E. ARCHER WARNE, President ROBERT JOHNSTON, Secretary E. K. SOPER, Faculty Sponsor EMILE HUGUENIN, Counselor Colorado School of Mines Golden, Colorado Student Chapter, A.I.M.E. J. Ross REED, President ANDREW PERRY, JR., Secretary J. BURNS READ, Faculty Sponsor A. W. NEWBERRY, Counselor

Jan 1, 1936

By D. Milne, G. Snell

"A new method to predict stope hangingwall failure in underground excavations is proposed. This method compares movement rate for stope hangingwalls, determined from changes in strain or deformation, with changing opening geometry. This geometry is represented by the effective radius factor (ERF), a measure of relative distance to supporting abutments from an opening surface point. This strain-ERF (SERF) approach is applied to an instrumentation program for Brunswick mine near Bathurst, New Brunswick. The SERF method for this case study predicted failure at an ERF of 13.9 m and hangingwall failure actually occurred at an approximate ERF of 12.4 m.RÉSUMÉ Cet article propose une nouvelle méthode permettant de prévoir la rupture de l’éponte supérieure des chambres d’abattage dans les excavations souterraines. Cette méthode compare la vitesse de mouvement des épontes supérieures des chambres d’abattage, déterminée à partir des changements aux niveaux des contraintes ou de la déformation, à la géométrie variable de l’ouverture au jour. Cette géométrie est représentée par le facteur du rayon effectif (ERF, de l’anglais Effective Radius Factor), une mesure de la distance relative jusqu’aux appuis latéraux de soutènement à partir d’un point de la surface d’ouverture au jour. Cette approche contraintes-ERF (SERF, de l’anglais strain-ERF) est appliquée à un programme d’instrumentation pour la mine Brunswick située près de Bathurst, dans le Nouveau-Brunswick. Pour cette étude de cas, la méthode SERF prévoyait une rupture à un ERF de 13,9 mètres (m); la rupture de l’éponte supérieure s’est effectivement produite à un ERF d’environ 12,4 m.INTRODUCTIONUnderground stope design can be improved with the installation of field instrumentation. The most commonly used instrumentation for monitoring rock mass behaviour around open stopes is extensometers. The y-axis of Figure 1 represents how the changing rate of movement, measured using either displacement- or stress-type instruments, can be observed over time. The magnitude of movement measured by extensometers can give an approximation of the load on installed support (Milne, Pakalnis, & Lunder, 1996) and a changing rate of movement can indicate approaching instability."

Jan 1, 2018

By J. Miettunen

"The lining of a grinding mill has an important role in creating the conditions for the grinding process. The main purpose of the lining is to transfer the energy taken by the mill motor to the contents of a mill. The power draw curve of the mill, if measured with sufficiently short intervals, describes how well the lining works in lifting and mixing the material to be ground.The power draw curve can also be used for determining the volumetric filling of a mill. This new measuring technique was developed at the Pyhasalmi concentrator in 1989-90. The measuring device with its PC-connections offers an excellent tool for controlling volumetric mill filling, for maximizing throughput and for research purpose (e.g. process studies, mill linings design).The paper describes the new method and the results achieved at Outokumpu's concentrators.INTRODUCTIONVarious methods have been developed for measuring mill load; weighing, noise, etc. Weighing is one of the most reliable methods but installing the sensor to existing mills is often difficult. Besides the load measurement does not indicate the volumetric filling if the specific gravity of the mill feed fluctuates. Noise measurement has often failed with its sensitivity to various disturbance sources.As known, the power draw measurement is one of the most accurate measurements in grinding circuits and therefore it has been utilized in sophisticated control, e.g. in power peak seeking algorithms for maintaining maximum grinding efficiency.About 10 years ago power oscillation measurement was developed at Pyhasalmi concentrator. In this method the amplitude of the oscillation caused by the lifters was measured. The oscillation amplitude and the mill filling has quite a good correlation and together with the power draw measurement it could be used for maximizing mill throughput (Koivistoinen and Miettunen, 1989). The main difficulties in this method were that the dependence between mill filling and the amplitude is unlinear and that the dependence also is individual for each mill.The method developed during 1989-90 is also based on the power oscillation but on its phase shift. In principle level of the material in a mill is measured. The method has been successfully used for maximizing throughput at Pyhasalmi and Enonkoski concentrators and at Harjavalta slag concentrator."

Jan 1, 1991

All of the coal resource in North Dakota is lignite. The lignite fields of North Dakota cover approximately 40 percent of the total area of the state, with reserves estimated to be 9.8 billion tons, two percent of the U.S. base. The first commercial coal mine in North Dakota opened in Morton County in 1873. As the railroads crossed the plains, the demand for coal increased. Coal for fuel was supplied by the state&apos;s underground mines. By 1884, North Dakota&apos;s lignite production had reached 35,000 tons. North Dakota was among the fist states to shift from underground to large-scale commercial surface coal mining. By 1927,40 percent of the total production was by surface mining, compared to 2 percent for the nation as a whole. By 1959,86 percent of the total production was by surface mining, while the national average was 22 percent. Since 1966, surface mining has been the exclusive coal production method in North Dakota. North Dakota&apos;s regulatory program received conditional approval December 15, 1980. Following approval of amendments to the program, it received full approval February 18, 1986. The Abandoned Mine Lands Program was approved December 23,1981.

Jan 1, 1992

By J. Faitli, I. Tarján

One of the basic processes of process engineering is the production of dense suspensions. The major two equipment of this task are the periodically operating propeller mixer, and the continuously operating mixer tank based on the flow principle. In the laboratory of the Department of Process Engineering at the University of Miskolc, laboratory scale experimental setups had been built of both types of mixers in the frame of the FKFP 0504/1999 signed research project, supported by the Hungarian Ministry of Education. By these experimental setups many systematical experimental series had been carried out by test materials of water as the dispersing media and sand, power plant fly ash and magnetite as the granular solids. On the basis of the literature and these experimental investigations, scale-up methods had been developed for both types of mixers for the determination of the main technical parameters of industrial size equipment.

Jan 1, 2003

By Peter Raziszewski

"A ball charge dynamics model is presented. Simulation results are compared with observed data showing good correlation. Based on the charge dynamics models, a liner wear model is presented along with some wear simulation results. Using the charge dynamics model with the liner wear model, it is possible to envision the predictions of mill power consumption as a function of liner wear. Further work is needed to validate the liner wear model.IntroductionFrom a process control point of view, industrial systems can be considered stationary or non-stationary. For example, the physical properties, such as geometric and/or material, of a system considered stationary do not change with time. However, this is not true for non-stationary systems such as those found in the mining industry where process dynamics are often violent with wear being an intrinsic process phenomenon.In this paper, a model for ball charge dynamics will be presented along with a description for liner wear. Simulation results for power draft prediction using the charge model will be compared with observed data. Further, liner wear results will also be presented."

Jan 1, 1997

By Willard P. Ward

The numerous direct processes which have been patented and brought before the iron-masters of the world, differ materially from that now introduced by Mr. Wilson. After a careful examination of his process, I am convinced that Mr. Wilson has succeeded in producing good blooms from iron-ore, and I think that I am able to point out theoretically the chief reasons of the success of his method. Without going deeply into the history of the metal, I may mention the well-known fact, that wrought-iron was extensively used in almost all quarters of the globe, before pig or cast-iron was ever produced. Without entering into the details of the processes by which this wrought-iron was made, it suffices for my present purpose to say that they were crude, wasteful, and expensive, so that they can IE employed to-day only in a very few localities favored with good and cheap ore, fuel, and labor. The constructiou of larger furnaces and the employment of higher temperatures led to the production of a highly carbonized, fusible metal, without any special design on the part of the manufacturers in producing it. This pig-iron, however, could be used only for a few purposes for which metallie iron was needed; but it was produced cheaply and with little loss of metal, and the attempt to decarbonize this product and bring it into a state in which it could be hammered and welded was soon successfully made. This process of decarbonizatioa, or some modification of it, has successfully held the field against all, so-called, direct processes up to the present time. Why ? Because the old-fashioned bloomeries and Catalan forges could produce blooms only at a high cost, and because the new processes introduced failed to turn out good blooms. Those produced were invariably "red-short," that is, they contained unreduced oxide of iron, which prevented the contact of the metallic particles, and rendered the welding together of these particles to form a solid bloom impossible. The process of puddling cast-iron, and transforming it by decarhonization into wrought-iron has, as everybody knows, been in successful practical operation for many years, and the direct process referred to so closely resembles this, that a short description of the theory of puddling is not out of place here.

Jan 1, 1884