3,4-Di-O-acetyl-~-xylal reacted with carbon n~onoxide and hydrogen in the presence of clicobalt octacarbonyl to yield 1,5-anhydro-4-deoxy-~-arab~)zo-hexitol (I) ancl 1,s-anhydro-4-deoxy-L-sylo-hexitol (11). The stereochemistry a t C-5 of the hesitols was elucidatecl by correlation with the triol ether (VI) obtained by sodi~~iil borohydride reduction of the perioclate oxidation product of 1,4~anhytlro-5-cleosy-~-arabi?~o-hexitol (VII). Reaction of 3,4-di-0-acet~l-D-xylal with carbon monoside and deuteri~~rll afforded 1,5-anhydro-4-cleosy-~-n1~1bit1o-hc~itol-4,6,6-H3~ (VIII) and 1,5-anhydr0-4-deoxy-~-.vylo-hexitol-4,,6-H ( I S ) . Examination of the nuclear magnetic resonance (n.m.r.) spectra of the nor~ual and cleuterated anhydrodeosyhexitols confirmed the structural assignments and shorvcd that cis-addition to the double bond takes place to give (IX). (3) to yield an anhydrodeoxpheptitol of unltnown stereochemistry. I t has now been found that pentals react similarly t o yield ailhydrodeoxyhexitols in which a hydroxymethyl group has been added to C-1 of the pental. The present paper in the series deals with the 0x0 reaction of 3,4-di-0-acetyl-~-x).lal and offers conclusive proof by chemical ineans of the stereochemistry of the hexitols obtained. In addition, the mechanism of the addition of carbon monoxide and hydrogen t o the glycal has been deterrnined by the use of tracer studies.The 3,4-di-O-acetyl-~-xylal, prepared by a nlodification of the procedure of Helferich and co-worliers (-I), was reacted with a mixture of carbon monoxide and hydrogen a t a

2,3-Di-O-acetyl-l,5-anhydro-4,6-dideoxy-~-nrabi?~o-hexitol (VI) and 2,3-di-0-acetyl-1,s-anhydro-4,6-dideoxy-I--rylo-hexitol (VII) were prepared by reaction of ~ilethyl magnesium bromide with 3,4-di-0-acetyl-~-xylo-pyra1losyl chloride. An independent synthesis of compounds (VI) and (VII) was also achieved from the 6-0-tosyl derivatives of the 0x0 products from 3,4-di-O-acetyl-~-xylal, thus confirming that the configuration of the secondary hydroxyls of glycals is not inverted during the 0x0 reaction.Part I1 of this series ( I ) showed that application of the 0x0 reaction to 3,4-di-0-acetyl-11-xylal, followed by deacetylation, gave 1,s-anhydro-4-deoxjr-D-arabino-hexitol (I) and 1 ,5-anhydro-4-deoxy-L-xylo-hexitol (I I). The I--configuration of C-5 of (I I) was established by correlation with C-4 of 1,4-anhydro-5-deoxy-D-arabino-hexitol; the C-5 configuration of (I) must therefore be D , a s this anhydrodeoxyhexitol, on cleavage of the 0-glycol bond and reduction, formed a trio1 ether which was the mirror image of that derived from (11). The remaining asymmetric centers of (I) and (11), a t C-2 and C-3, were assuined to be unchanged in configuration from the parent polyol.Von Rudloff et al. (2) carried out the hydrogenolysis of methyl-a-D-glu~op~ranoside a t 210" t o give 1,s-anhydro-4-deoxy-hexitol along with other products. Gorin (3) subsequently sho\\red that the hexitol consisted of two 1,s-anhydro-1-deoxy-D-hexitols. On the bases of coillparative rates of lead tetracetate oxidation and con~parison of observed specific rotations with those calculated from application of the principle of opt-ical superposition, these hexitols were postulated to have structures (I) and (111) (1,;S-anhydro-4-deoxy-D-lyxo-hexitol). The compound t o which they assigned the D-nrabino configuration +19"; tris-p-nitrobenzoate, 111.p. 115-119", [a]D +59") is clearly different from our con~pound (I) ([DID -13"; tris-p-nitrobenzoate, 1n.p. 21;3", [elD -50"). We have accordingly carried out a series of reactions correlating the secondary hydroxyl configurations of (I) and (11) with those of ~-x y l a l , thereby confirming our assignments of structures (I) and (11). Bonner (1) has shown that the hemiacetal halogen of pol yacetylglycos>~l halides undergoes a et eta the tical reaction with Grignard reagent \\.hen the latter is present in sufficient-excess, and has, moreover, provided adecl~~ate experirnental proof to sho\\-that decomposition of the Grignard adduct t o the ester group does not involve rupture of asymmetric bonds, and therefore no change of configuration talres place during reaction with the Grignard reagent. lising the Grignard synthesis ure have been successful in preparing the diacetates of the two 6-deoxy analogues of (I) and (11) For personal use only.

Hydroformylation of 3,4-di-0-acetyl-~-xylal gave 4,5-di-0-acetyl-2,6-anhydro-3-deoxyaldehyde-D-lyxo-hexose (I) and 4,5-di-0-acetyl-2,6-a1~hydro-3-deoxy-uldehydo-~-xylo-hexose iII). llethyl sulfoxide oxidation of 2,3-di-0-acet~~1-1,5-anhydro-4-deoxy-~-u~ubino-hexitol (111) and of 2,3-di-0-acetyl-l,5-anhl-dro-L-xylo-hex (IV) yielded compoullds I and 11, respectively.Reaction of 3,4,6-tri-0-acetyl-D-glucal with carbon inonoxide and hydrogen under carefully controlled conditions yielded 4,5,7-tri-0-acetyl-2,6-anhydro-3-deoxy-uldehydo-~-~unno-heptose (VII) and 4 , . 5 , 7 -t r i -0 -a c e t y l -2 , 6 -a n h y d r o -3 -d e o x~c o -h e p t o s e (VIII) in a 707; yield. Compound V I I was readily separated from the mixture of aldoses and alditols by column chromatography. Alternatively, aldoses VII and VIII were separated via their P,4-dinitrophenylhydrazones. The co~lversio~l of aldoses \;I1 and VIII into their diethyl dithioacetal derivatives is also described.The fortnyl group located in an equatorial orientation was found t o be reduced a t a faster rate than its axial counterpart.

The reaction of 3>4-di-0-acetyl>i|D-xylal with 3 moles of synthesis gas (CO + 2Hg) under oxo conditions gave predominantly two isomeric 2,3-di-0-acetyl-l,5-anhydro-4-deoxy hexitols, by addition of a hydroxymethyl group at C-l of the glycal. The structures of the two polyols, obtained by deaeetylation of the reaction product and fractionation by paper partition chromatography, were completely established. Formation of a pair of enantiomeric triol ethers by periodate cleavage and sodium borohydride reduction of each polyol showed that they were l,5-anhydro-4-deoxyhexitols, having unbranched carbon skeletons, this also being shown by the proton resonance positions and intensities in the n.m.r. spectra of the polyols. One of the enantiomeric triol ethers, having the L-configuration, was prepared from a carbohydrate of known structure, 1,4-anhydro-5-deoxy-B-arabino-hex it o1,.thereby establishing the configurations at C-5 of the two isomeric 1,5anhydro-4-deoxy-hexitols. Assignments of the D-arabinoand L-xylo-configurations to the two isomers conflicted " ~~ ftp with results of Qorin , who had previously assigned the D-arablno-configurations to a 1,5-anhydro-4-deoxy-hexi161 which did not resemble either of our compounds. That these were the D-arabino-and,L-xylo-isomers of 1,5-anhydro-4-iv deoxy-hexitol was proved by their conversion into a pair of isomeric l,5-anhydro-4,6-dideoxy-hexitols which were identical with those obtained by the reaction of 3,4-di-Oacetyl-2-deoxy-D-xylopyranosyl chloride with methyl magnesium bromide, both series of reactions allowing no possibility of configurational inversions. The polyol described by Gorin was subsequently shown to be the alternative trans isomer, 1,5-anhydro-4-deoxy-D-xylo-hexitol. A concurrent study of the structures of the two anhydrodeoxyhexitols was made by nuclear magnetic resonance, and the stereochemistry of the L-xylo-isomer could be assigned from the multiplicities of the C-4 proton signals. The single C-4 proton in the deuterated analogue of the L-xylo-isomer (prepared by reacting 3,4-di-O-acetyl-Dxylal with carbon monoxide and deuterium) was shown to be equatorial by its resonance position, and its multiplicity on deuterium-hydrogen decoupling, this providing evidence for cis-addition to the double bond of the glyeal on hydroformylation. The oxo reaction of 3,4,6-triO -acetyl-D-galactal has been reinvestigated, and found to be entirely analogous to those of other glycals, giving, on deacetylation, a mixture of 2,6-anhydro-3-deoxy-D-galactoand D-taloheptitols. These were isolated and characterised, and their V stereochemistry established by correlation with the D-glucoisomer, whose structure has been proved by X-ray analysis. The reaction of 3,4-di-O-acetyl-D-xylal under hydroformylation conditions, leading to the formation of aldehydes rather than alcohols, has been investigated. From the reaction of the glycal with 2 moles of synthesis gas, two isomeric 4,5-di-0-acety1-2,6-anhydro-3-deoxy-aldehydohexoses were isolated as their crystalline 2,4-dinitrophenylhydrazo...

o-Hydroxybenzaldehyde and some of its derivatives react with two molecules of certain diphenylenes in acetic acid saturated with dry hydrogen chloride to give benzocyclopentafuran derivatives and not chromens as indicated in related reactions described by Wizinger,l who used perchloric acid in acetic anhydride-acetic acid.

3,4,6-Tri-0-acetyl-D-galactal reacted with carbon monoxide and hydrogen in the presence of dicobalt octacarbonyl t o yield 2,6-anhydro-3-deoxy-D-galado-heptitol (I) and 2,6-anhydro-3-deoxy-D-lalo-heptitol (11). The stereochemistry of (I) and (11) has been established by correlation with 2,6-anhydro-3-deoxy-D-gluco-heptitol (VI) of lcnown configuration. Evidence that both (I) and (11) are formed by the addition of a hydroxmethyl group t o C-1 of the glycal is also furnished by the proton 1l.m.r. spectra of the normal and deuterated anhydrodeoxyheptitols.Part I of this series (1) described the action of carbon monoxide and hydrogen on 3,4,6-tri-0-acetyl-~-galactal, and presented evidence for the formation of a n anhydrodeoxyheptitol of unknown stereochemistry. I t has since been found t h a t 3,4,6-tri-0-acetyl-D-galactal reacts under 0x0 conditions in a manner entirely analogous t o other glycals investigated (2,3) to yield a s the major products a pair of isomeric anhydrodeoxyheptitols (I) and (11), formed by the addition of a hydroxymethyl group to C-1 of the unsaturated carbohydrate. The stereochemistry of (I) and (11) has been elucidated by a n analysis of their proton n.m.r. spectra and by correlation with 2,6-anhydro-3-deoxy-D-gbco-heptitol (VI), whose structure has been proved by crystallographic X-ray analysis (4).3,4,6-Tri-0-acetyl-D-galactal (5) was reacted in a high-pressure apparatus with carbon monoxide and hydrogen in the presence of dicobalt octacarbonyl, under conditions similar to those described previously (1). The reaction product was separated from the catalyst by chromatography on a Florisil column, and deacetylated with methanolic sodium methoxide. The two major reaction products (I) and (11), after separation by preparative paper chromatography, were obtained in approximately equal amounts. DISCUSSION Both (I) and (11) consumed 1 molar equivalent of periodate (0.95 and 0.98 respectively) as measured by the spectrophotometric method of Dixon and Lipkin (6). Fraction (11) reacted with acetone in the presence of sulfuric acid to form a monoisopropylidene derivative (VII), which on treatment with p-toluenesulfonyl chloride in pyridine gave a compound (VIII) containing two tosyloxy groups. The latter compound on heating with sodiunl iodide in acetone solution liberated 2 equivalents of sodium p-toluenesulfonate. Thus i t is highly likely t h a t (11) contains two primary hydroxyl groups and two adjacent cis secondary hydroxyl groups. Compound (I) was characterized as the tetra-0-(p-nitrobenzoyl) derivative.Evidence that both (I) and (11) have unbranched carbon skeletons was furnished by their proton nuclear magnetic resonance (n.m.r.) spectra, measured in DzO solution (Fig. l a and c). These both show a group of signals a t lower field (3.3-4.3 p.p.m.) with total area corresponding t o 8 hydrogens, and a further group a t higher field (1.3-2.3 p.p.m.), area = 2 hydrogens. The higher field signals can clearly be assigned t o hydrogens attached